View ISD-T360SB_2066673.PDF datasheet online --- IC-ON-LINE

Datasheet File OCR Text:

table of contents isd vii chapter 1?hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1 1.1 pin assignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1 1.1.1 pin-signal assignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1 1.2 description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3 1.2.1 resetting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3 1.2.2 clocking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3 1.2.3 power-down mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-4 1.2.4 power and grounding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-5 1.2.5 memory interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-6 1.2.6 the codec interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-13 1.3 specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-18 1.3.1 absolute maximum ratings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-18 1.3.2 electrical characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-18 1.3.3 switching characteristics?reliminary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-20 1.3.4 synchronous timing tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-23 1.3.5 timing diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-25 chapter 2?software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1 2.1 system operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1 2.1.1 the state machine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1 2.1.2 command execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2 2.1.3 event handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2 2.1.4 message handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-3 2.1.5 tone generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-4 2.1.6 initialization and configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-4 2.1.7 power-down mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-5 2.2 peripherals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-5 2.2.1 microcontroller interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-5 2.2.2 memory interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-8 2.2.3 codec interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9 2.3 algorithm features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9 2.3.1 vcd (voice compression and decompression) . . . . . . . . . . . . . . . . . . . . . . . 2-10 2.3.2 dtmf detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-11 2.3.3 tone and energy detection (call progress) . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-12
ISD-T360SB viii voice solutions in silicon 2.3.4 full-duplex speakerphone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-14 2.3.5 speech synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-16 2.4 voicedsp processor commands?uick reference table . . . . . . . . .2-21 2.5 command description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2-24 chapter 3?schematic diagrams . . . . . . . . . . . . . . . . . . . 3-1 3.1 application information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-1 chapter 4?physical dimensions . . . . . . . . . . . . . . . . . . . 4-1
isd ?2045 hamilton avenue, san jose, ca 95125 ?tel: 408/369-2400 ?fax: 408/369-2422 ?http://www.isd.com december 1998 ?dvance information ISD-T360SB voicedsp digital speech processor with master/slave, full-duplex speakerphone, multiple flash and aram/dram support the voicedsp product family combines multi- ple digital signal processing functions on a single chip for cost-effective solutions in telephony, au- tomotive and consumer applications. the voicedsp processor offers necessary fea- tures to modern telephony products, such as: high-quality, speech record and playback, elec- trical and acoustic echo cancellation for full-du- plex hands-free speakerphone operation. the ISD-T360SB voicedsp can be used in various applications: digital telephony with add-on speech processing: digital telephone answering machines(dtads); and full-duplex, hands- free speakerphone operation for isdn, dect, digital spread spectrum, and analog cordless applications; ct0/1 base stations. an add-on chip for corded telephones featuring dtad functions and/or full- duplex, hands-free speakerphone operation. stand-alone digital answering machines with full-duplex, hands-free speakerphone operation. voice memo recording automotive applications employing full- duplex speakerphone operations for hands-free, in-car communications, and for car status and information announcements. based on isd? unique concept which combines 16-bit dsp (digital speech processor) and 16-bit risc core technology, the ISD-T360SB is a high- performing chip solution for various applications. to facilitate incorporating the voicedsp proces- sor, it features system support functions such as an interrupt control unit, codec interface (mas- ter, slave), microwire interface to the system mi- crocontroller, as well as a memory handler for flash and dram memory devices. design of high-end, price optimized systems are possible with isd? voicedsp flexible system interfaces (codec, microcontroller, and memory manage- ment support). the ISD-T360SB processor operates as a peripher- al controlled by the system microcontroller via an enhanced, serial microwire interface. the sys- tem microcontroller typically controls the analog circuits, buttons and display, as well as activates functions through commands. the voicedsp ex- ecutes these commands and returns status infor- mation to the microcontroller. the voicedsp software resides in the on-chip rom. it includes dsp-based algorithms, system support functions, and a software interface to hardware peripherals. preliminary information
ISD-T360SB ii voice solutions in silicon features at a glance dtad manag e ment highest quality speech recording in pcm format for music on hold or other ogm (out going message) recording and ivs selectable high-quality speech compression rate of 5.3 kbit/s, 9.9 kbits or 16.8 kbit/s, plus silence compression with each rate up to 16 minutes recording on a 4-mbit flash high-quality music compression for music on hold (16.8 kbits) programmable message tag for message categorization, e.g., mailboxes, incoming messages (icm), outgoing messages (ogm) message management skip forward or backward during message playback variable speed playback real-time clock: day of week, hours, minutes multi-lingual speech synthesis using international vocabulary support (ivs) vocabularies available in: english, japanese, mandarin, german, french and spanish software automatic gain control (agc) speakerphone digital full-duplex speakerphone acoustic- and line-echo cancellation continuous on-the-fly monitoring of external and internal conditions (acoustic and line) provides high-quality, hands-free, conversation in a changing environment minimum microcontroller control intervention (launch-and-forget) supports: on, off, mute, and hold functions call and device management digital volume control least cost routing support (lcr) power-down mode 3v to 5v selectable power supply 4.096 mhz operation dtmf generation and detection telephone line functions, including busy and dial tone detection single tone generation dtmf detection during message playback peripheral control codec ?law, a-law, and 16-bit linear codec input support selectable master/slave codec interface supports two in-coming lines in slave mode without speakerphone for dtad recording supports up to 16 user selectable speech channels in slave mode supports long-frame and short-frame codecs single/double bit clock rate for slave mode on-chip codec clock generation memory supports up to four 4-mbit, four 8-mbit, or four 16-mbit flash devices from toshiba or samsung supports up to four 16-mbit aram/dram memory devices from toshiba, samsung, and samsung-compatible the number of messages that can be stored is limited only by memory size direct access to message memory
ISD-T360SB iii isd message storage contains all data in a concatenated chain of memory blocks. memory mapping and product floor test included supports external vocabularies, using flash memory or expansion rom microwire microwire slave interface to an external microcontroller sophisticated command language to optimize system code size international vocabulary support (ivs) for manufacturing recorded voice prompt and speech synthesis, the isd international vocabu- lary support delivers pre-recorded voice prompts in the same high-quality of the user-re- corded speech. for complete control over qual- ity and memory management, the ivs features adjustable speech compressions. in addition, several pre-recorded voice prompt sets are available in various languages for further conve- nience. available languages: english japanese mandarin german french spanish develop a new vocabulary by isd? voice prompt development tool, the isd-ivs360. this vocabulary development tool supports various languages and including their unique grammar structures. isd-ivs360, pc-windows95-based program, synthesizes recorded .wav files into the ISD-T360SB? various compression rates (including pcm). isd? voicedsp products store ivs vocabularies on either flash memory or expansion rom mem- ories, thus dtad manufacturers can design a product for multiple countries, featuring various languages. for more details about ivs, refer to the ivs user? guide .
ISD-T360SB iv voice solutions in silicon figure 1-1: ISD-T360SB block diagram?asic configuration with four 4mb/8mb/16mb nand flash devices (samsung/toshiba)
ISD-T360SB v isd figure 1-2: ISD-T360SB block diagram?asic configuration with four 4mb serial toshiba flash devices
ISD-T360SB vi voice solutions in silicon figure 1-3: ISD-T360SB block diagram?asic configuration with four 16mb aram/dram devices (samsung) and ivs eprom
1?ardware ISD-T360SB 1-1 isd chapter 1?hardware 1.1 pin assignment the following sections detail the pins of the isd- t360sb processor. slashes separate the names of signals that share the same pin. 1.1.1 pin-signal assignment table 1-1 shows all the pins, and the signals that use them in different configurations. it also shows the type and direction of each signal. figure 1-1: 80-mqfp package connection diagram 60 59 58 57 56 55 54 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 nc nc nc mwrdy v ss mwdout v cc mwclk nc d1 d2 v ss d5 v ss cas /mmclk v cc a6 a5 a4 a3 a2 v cc a1 nc nc nc pb7 pb6 pb5 pb4 pb3 v ss pb2 v cc pb1 pb0 pc7 pc6/emcs /env0 pc5 pc4/a15 pc3/a14 top view 80-mqfp 18 19 20 40 39 38 reset 53 52 51 50 49 48 47 46 45 44 43 42 41 mwdin cclk cdin cfs0 61 62 63 nc nc a10 v cc hi v cc d4 d6 d7 pc0/a11 nc nc v ss a7 v cc a x2/clkin x1 tst nc mwrqst mwcs pc2/a13 nc pc1/a12 nc d3 a9 d0 dwe /mmdin ras /mmdout ISD-T360SB a0 a8 v ss a cdout cfs1
1?ardware ISD-T360SB 1-2 voice solutions in silicon 1. ttl1 output signals provide cmos levels in the steady state, for small loads. 2. input during reset. cmos level input. 3. virtual address lines for ivs rom. 4. chip select lines for serial flash devices. 5. schmitt trigger input. table 1-1: voicedsp pin signal assignment pin name signal name type description a(0:15) a(0:16) output address bits 0 through 16 cas cas output dram column address strobe cclk cclk i/o codec master/slave clock cdin cdin input data input from codec cdout cdout output data output to codec cfs0 cfs0 i/o codec 0 frame synchronization cfs1 cfs1 output codec 1 frame synchronization d(0:7) d(0:7) i/o data bits 0 through 7 dwe dwe output dram write enable emcs /env0 emcs output expansion memory chip select emcs /env0 env0 input environment select mmclk mmclk output master microwire clock mmdin mmdin input master microwire data input mmdout mmdout output master microwire data output mwclk mwclk input microwire clock mwcs mwcs input microwire chip select mwdin mwdin input microwire data input mwdout mwdout output microwire data output mwrdy mwrdy output microwire ready mwrqst mwrqst output microwire request signal pb(0:7) 3 pb(0:7) i/o port b, bits 0 through 7 pc(0:7) pb(0:7) i/o port c, bits 0 through 7 ras ras output dram row address strobe reset reset input reset tst tst input test pin v cc v cc power 3.3 v power supply pin v cc av cc a power 3.3 v analog circuitry power supply pin v cc hi v cc hi power 5 v power supply pin. connect to v cc if 3.3 v power supply is used. v ss v ss power ground for on-chip logic and output drivers v ss av ss a power ground for on-chip analog circuitry x1 x1 oscillator crystal oscillator interface x2/clkin x2 oscillator crystal oscillator interface
1?ardware ISD-T360SB 1-3 isd 1.2 description this section provides details of the functional characteristics of the voicedsp processor. it is di- vided into the following sections: resetting clocking power-down mode power and grounding memory interface codec interface 1.2.1 resetting the reset pin is used to reset the voicedsp pro- cessor. on application of power, reset must be held low for at least t pwr after v cc is stable. this ensures that all on-chip voltages are completely stable before operation. whenever reset is applied, it must also remain active for not less than t rst , see table 1-10 and table 1-11. during this period, and for 100 ms after, the tst signal must be high. this can be done with a pull-up resistor on the tst pin. the value of mwrdy is undefined during the re- set period, and for 100 ms after. the microcon- troller should either wait before polling the signal for the first time, or the signal should be pulled high during this period. upon reset, the env0 signal is sampled to deter- mine the operating environment. during reset, the emcs /env0 pin is used for the env0 input sig- nals. an internal pull-up resistor sets env0 to 1. after reset, the same pin is used for emcs . system load on env0 for any load on the env0 pin, the voltage should not drop below v envh (see table 1-10 and table 1-11). if the load on the env0 pin causes the current to exceed 10 ?, use an external pull-up resistor to keep the pin at 1. figure 1-2 shows a recommended circuit for generating a reset signal when the power is turned on. figure 1-2: recommended power-on reset circuit 1.2.2 clocking the voicedsp processor provides an internal os- cillator that interacts with an external clock source through the x1 and x2/clkin pins. either an external single-phase clock signal, or a crystal oscillator, may be used as the clock source. external single-phase clock signal if an external single-phase clock source is used, it should be connected to the clkin signal as shown in figure 1-3, and should conform to the voltage-level requirements for clkin stated in ?lectrical characteristics?on page 1-18. v cc reset isd-t360 v cc v ss
1?ardware ISD-T360SB 1-4 voice solutions in silicon figure 1-3: external clock source crystal oscillator a crystal oscillator is connected to the on-chip oscillator circuit via the x1 and x2 signals, as shown in figure 1-4. figure 1-4: connections for an external crystal oscillator keep stray capacitance and inductance, in the oscillator circuit, as low as possible. the crystal resonator, and the external components, should be as close to the x1 and x2/clkin pins as possi- ble, to keep the trace lengths in the printed cir- cuit to an absolute minimum. you can use crystal oscillators with maximum load capacitance of 20 pf, although the oscilla- tion frequency may differ from the crystal? spec- ified value. table 1-2 lists the components in the crystal oscil- lator circuit 1.2.3 power-down mode power-down mode is useful during a power fail- ure or in a power-saving model, when the power source for the processor is a backup battery or in battery-powered devices, while the processor is in idle mode. in power-down mode, the clock frequency of the voicedsp processor is reduced and some of the processor modules are deactivated. as a re- sult, the ISD-T360SB consumes considerably less power than in normal-power mode. although the voicedsp processor does not perform all its usual functions in power-down mode, it does re- tain stored messages and maintain the date and time. note in power-down mode all the chip select signals, cs0 to cs3, are set to 1. to guaran- tee that there is no current flow from these signals to the flash devices, the power sup- ply to these devices must not be disconnected. the ISD-T360SB stores messages and all memory management information in flash or aram/ dram memory. when flash memory is used for memory management, power does not need to be maintained to the processor to preserve stored messages. when aram/dram memory is used for message management, preserving stored messages requires a battery back up dur- x2/clkin clock generator x1 voicedsp single-phase clock signal x1 x2 c1 r1 c2 isd-t360 table 1-2: components of crystal oscillator circuit component values tolerance crystal resonator 4.096 mhz resistor r1 10 m w 5% capacitors c1, c2 33 pf 20%
1?ardware ISD-T360SB 1-5 isd ing a power failure. if the power failure extends the life of the battery, the microcontroller should perform the initialization sequence (as described on page 2-4), and use the setd command to set the date and time. to keep power consumption low during power- down mode, the reset , mwcs , mwclk and mwdin signals should be held above v cc ?0.5 v or below v ss + 0.5 v. 1.2.4 power and grounding power pin connections the isd-t360 can operate over two supply volt- age ranges 3.3 v ?0% and 5 v ?0%. the five power supply pins (v cc , v ss , v cc a, v ss a and v cc hi) must be connected as shown in figure 1-5 when operating in a 3.3 v environ- ment, and as shown in figure 1-6 when operat- ing in a 5 v environment. failure to correctly connect the pins may result in damage to the device. the capacitor and resistor values are given in table 1-3. 1. all capacitors represent two parallel capacitors at the values 1.0 ? and 0.1 ?. figure 1-5: 3.3 v power connection diagram table 1-3: components of supply circuit component values tolerance resistor r1, r2 10 w 5% capacitors (1) c1, c2 capacitors c3, c4, c5, c6, c7 1.0 ? tantalum 0.1 ? ceramic 20% 74 72 64 63 61 60 50 48 9 11 12 30 32 v ss a c 1 v ss v cc v cc hi v ss v cc v cc v ss 3.3 v supply isd-t360 v ss v cc av ss v cc v cc r 1
1?ardware ISD-T360SB 1-6 voice solutions in silicon figure 1-6: 5 v power connection diagram for optimal noise immunity, the power and ground pins should be connected to v cc and the ground planes, respectively, on the printed circuit board. if v cc and the ground planes are not used, single conductors should be run direct- ly from each v cc pin to a power point, and from each gnd pin to a ground point. avoid daisy- chained connections. the voicedsp does not perform its usual functions in power-down mode but it still preserves stored messages, maintains the time of day and generates aram/dram re- fresh cycles. when you build a prototype, using wire-wrap or other methods, solder the capacitors directly to the power pins of the voicedsp processor sock- et, or as close as possible, with very short leads. 1.2.5 memory interface flash support the ISD-T360SB voicedsp supports flash devices for storing recorded data, thus, power can be disconnected to the ISD-T360SB without losing data. the ISD-T360SB supports serial and semi- parallel flash device interfaces, such as tc58v16bft, tc5816bft, tc58a040f, km29n040t, km2928000t/it, and km29216000at/ait. the isd- t360sb may be connected to up to four flash de- vices, resulting with maximum recording storage of 16-mbits x 4 = 64 mbits (up to 4 hours of record- ing time). the following flash devices are supported: 74 72 64 63 61 60 50 48 9 11 12 30 32 v ss a c 7 v ss v cc v cc hi v ss v cc v cc v ss 5 v supply isd-t360 c 5 v ss v cc v cc a v cc v ss c 4 c 3 c 2 r 2 c 6
1?ardware ISD-T360SB 1-7 isd internal memory organization the flash devices detailed in table 1-4 divide in- ternally into basic 4-kbyte block units. the isd- t360sb uses one block on each device for mem- ory management, leaving the rest of the blocks available for recording. using at least one block for a single recorded message yields a maximum of num_of_blocks_in_mem ?1 (see table 2-10 for parameter definition) messages per device. upon initialization the ISD-T360SB activates a sift- ing algorithm to detect defected blocks. de- fected blocks are defined as blocks with over 10 bad nibbles (a nibble consists of 4 bits). the de- fected blocks are marked as unused and ex- cluded from the list of available blocks for recording. toshiba serial flash the voicedsp processor supports up to four tc58a040f, 4mbit, serial interface, flash memory devices for storing messages. the tc58a040f is organized as an array of 128 blocks, with a ded- icated, read-only, bad block map programmed by the manufacturer and located in the last block. this map is used by the ISD-T360SB to de- fine the available blocks for recording. the isd- t360sb uses the voicedsp master microwire in- terface to communicate, serially, with the flash devices, while selecting the current flash device using pb3-pb6. connecting less than four flash devices require connecting the flash devices se- quentially, starting from pb3 up to pb6 (see figure 1-7). refer to figure 1-34for the master mi- crowire timing diagram. table 1-4: supported flash devices manufacturer memory device name characteristics operating voltage memory size conversion type toshiba tc58v16bft 2mx8 3.3 v 16-mbit lv ?law toshiba tc5816bft 2mx8 5 v 16-mbit a-law toshiba tc58a040f serial 5v 4-mbit ?law, a-law samsung km29n040t 512kx8 5 v 4-mbit ?law, a-law, lv samsung km29w8000t/it 1mx8 5 v 8-mbit ?aw samsung km29w16000at /ait 2mx8 5 v 16-mbit a-law
1?ardware ISD-T360SB 1-8 voice solutions in silicon figure 1-7: memory interface with four toshiba 4mbit serial flash devices and optional voice ivs eprom
1?ardware ISD-T360SB 1-9 isd figure 1-8: memory interface with four 4mb/16 mbit, nand flash devices (samsung, toshiba) nand flash (samsung, toshiba) the voicedsp processor supports up to four, semi-parallel interface, flash memory devices for storing messages. flash device with semi-parallel interface uses a single 8bit i/o port to set the ad- dress and access the data. the ISD-T360SB sup- ports three types of flash volumes (4mbit, 8mbit and 16mbit as listed in table 1-4) while all the connected flash devices must be of the same type. ports b and c are used to connect isd- t360sb to the flash devices using port b for ad- dress and data transfer and port c for communi- cation control and chip select. connecting less than four flash devices require connecting the flash devices sequentially, starting from pc4 up to pc7 (see figure 1-8). the ISD-T360SB scans the flash devices upon initialization, sifting out the bad blocks, and marking them in a special map, located in the last block of each device.
1?ardware ISD-T360SB 1-10 voice solutions in silicon figure 1-9: memory interface with four toshiba 4mbit serial flash devices and optional ivs eprom flash endurance a flash memory may be erased a limited num- ber of times. to maximize the flash use, the mem- ory manager utilizes the flash? blocks evenly (i.e., each block is erased more or less the same number of times), to ensure that all blocks have the same lifetime. refer to the respective flash memory device data sheets for specific endur- ance specifications. a voicedsp processor message uses at least one block. the maximum recording time depends on four factors: 1. the basic compression rate (5.3 kbit/s, 9.9kbit/s, or 16.8kbit/s). 2. the amount of silence in the recorded speech. 3. the number of bad blocks.
1?ardware ISD-T360SB 1-11 isd the number of recorded messages. (the basic memory allocation unit for a message is a 4- kbyte block, which means that half a block on average is wasted per recorded message). aram/dram support the voicedsp processor supports up to four, 16-mbit, aram/dram devices for storing mes- sages. the isd-t360 connects to the aram/ dram device using address buses a0?11 and data buses d0?3. this connection allows ac- cess to 2 22 nibbles (16-mbit) on each device. the ISD-T360SB selects the current aram/dram de- vice using pb3?b6 as described in figure 1-10. using less than four aram/dram devices re- quires connecting the devices sequentially, starting from pb3 up to pc6. ras and cas are connected in parallel to all the aram/dram de- vices and are used to refresh the memory. the difference between aram and dram resides with the amount of bad cells on the device and the device performance. while dram has no bad cells, aram contains certain level of impuri- ty and thus requires testing and mapping of the bad blocks upon the initialization of the isd- t360sb. although there are no real blocks on the aram device, the ISD-T360SB emulates virtual ?locks?on the aram device (as if it was a flash device), tests these ?locks?and marks them in a special map on the last ?lock?on each de- vice. this test is required only when using aram devices (as opposed to dram devices that re- quire no testing due to lack of bad blocks). the virtual division to blocks simplifies the use of aram/dram devices and allows the use of the same set of commands for flash and aram/ dram. another major difference between aram/ dram and flash devices is that the internal map- ping in the aram is lost upon power off. thus, the initialization process needs to take place after each power reset. the mapping is not lost when entering and exiting the power-down mode. refer to figure 1-24 through figure 1-28 for timing diagrams of aram/dram read, write, refresh in normal mode and refresh in power-down mode cycles. rom interface ivs vocabularies can be stored in either flash memory and/or rom. the voicedsp processor supports ivs rom devices through an expansion memory mechanism. up to 64 kbytes (64k x 8) of expansion memory are directly supported. nev- ertheless, the processor uses bits of the on-chip port (pb) to further extend the 64 kbytes address space up to 0.5 mbytes address space. rom is connected to the voicedsp processor us- ing the data bus, d(0:7), the address bus, a(0:15), the extended address signals, ea(16:18), and expansion memory chip select, emcs , con- trols. the number of extended address pins to use may vary, depending on the size and config- uration of the rom. ISD-T360SB configured with semi-parallel flash memory can not support ex- tension rom. table 1-5: recording time with 15% silence compression memory size compression rate total recording time 4 mbit 5.3 kbit/s 14.9 minutes 4 mbit 9.9 kbit/s 8.1 minutes 4 mbit 16.8 kbit/s 4.8 minutes 8 mbit 5.3 kbit/s 29.8 minutes 8 mbit 9.9 kbit/s 16.2 minutes 8 mbit 16.8 kbit/s 9.6 minutes 16 mbit 5.3 kbit/s 59.6 minutes 16 mbit 9.9 kbit/s 32.4 minutes 16 mbit 16.8 kbit/s 19.2 minutes 32 mbit 5.3 kbit/s 119.2 minutes 32 mbit 9.9 kbit/s 64.8 minutes 32 mbit 16.8 kbit/s 38.4 minutes
1?ardware ISD-T360SB 1-12 voice solutions in silicon figure 1-10: memory interface with four 16-mbit aram/dram devices (samsung, toshiba) and optional ivs eprom table 1-6: supported dram devices manufacturer memory device name characteristics operating voltage memory size samsung km44c4004cs-6 edo 4mx4 5 v 16-mbit edo lv samsung km44v4004cs-6 edo 4mx4 3.3 v 16-mbit edo toshiba tp edo 4mx4 5v, 3.3v 16-mbit edo, lv
1?ardware ISD-T360SB 1-13 isd 1.2.6 the codec interface the isd-t360 provides an on chip interface for analog and digital telephony, supporting master and slave codec interface modes. in master mode, the isd-t360 controls the operation of the codec for use in analog telephony. in the slave mode, the isd-t360 codec interface is controlled by an external source. this mode is used in digital telephony (i.e., isdn or dect lines). the slave mode is implemented with respect to iom-2/ cgi specifications. see table 1-7 for codec options for the isd- t360sb (isd supports compatible codecs in addi- tion to those listed below). the codec interface supports the following fea- tures: master mode or slave mode. 8- or 16-bit channel width. long (variable) or short (fixed) frame protocol. single or double bit (slave mode only) clock rate. single or dual channel codecs one or two codecs multiple clock and sample rates. one or two frame sync signals this codec interface uses five signals: cdin, cd- out, cclk, cfs0, and cfs1. the cdin, cdout, cclk, and cfs0 pins are connected to the first codec. the second codec is connected to cdin, cdout, cclk, and cfs1 pins. data is transferred to the codec through the cdout output pin. data is read from the codec through the cdin input pin. the cclk and cfs0 pins are output in master mode and input in slave mode. the cfs1 is an output pin. short frame protocol when the short frame protocol is configured, eight or sixteen data bits are exchanged with each codec in each frame (i.e., the cfs0 cycle). data transfer begins when cfs0 is set to 1 for one cclk cycle. the data is then transmitted, bit by bit, via the cdout pin. concurrently, the re- ceived data is shifted in through the cdin pin. data is shifted one bit per cclk cycle. after the last bit has been shifted, cfs1 is set to 1 for one cclk cycle. then, the data from the second co- dec is shifted out via cdout, concurrently with the inward shift of the data received via cdin. long frame protocol when long frame protocol is configured, eight or sixteen data bits are exchanged with each co- dec, as for the short frame protocol. however, for the long frame protocol, data transfer starts by setting cfs0 to 1 for eight or sixteen cclk cy- cles. short or long frame protocol is available in both master and slave modes. figure 1-11 illus- trates an example of short frame protocol with an 8-bit channel width. table 1-7: supported codec devices manufacturer codec device name characteristics operating voltage conversion type national semiconductor tp3054 single codec 5 v ?law national semiconductor tp 3057 single codec 5 v a-law oki msm7533v dual codec 5 v ?law, a-law oki msm 7704 dual codec 3.3 v ?law, a-law, lv macronix mx93002fc dual codec 5 v ?law lucent t7502 dual codec 5 v a-law lucent t7503 dual codec 5 v ?law
1?ardware ISD-T360SB 1-14 voice solutions in silicon figure 1-11: codec protocol-short frame?-bit channel width channel width the codec interface supports both 8-bit and 16- bit channel width in master and slave modes. slave mode the voicedsp supports digital telephony appli- cations including dect and isdn by providing a slave mode of operation. in slave mode opera- tion, the cclk signal is input to the isd-t360 and controls the frequency of the codec interface operation. the cclk may take on any frequen- cy between 500 khz and 4 mhz. both long and short frame protocol are supported with only the cfs1 output signal width affected. the cfs0 in- put signal must be a minimum of one cclk cy- cle. in slave mode, a double clock bit rate feature is available as well. when the codec interface is configured to double clock bit rate, the cclk in- put signal is divided internally by two and the re- sulting clock used to control the frequency of the codec of the codec interface operation. this interface supports isdn protocol with one bit clock rate or double bit clock rate. the exact for- mat is selected with the cfg command. the slave codec interface uses four signals: cdin, cdout, cclk, and cfs0. the cdin, cclk, and cfs0 input pins and the cdout output pins are connected to the isdn/dect agent. data is transferred to the voicedsp through the cdin pin and read out through the cdout pin. the cfs0 pin is used to define the start of each frame (see below) the source of that signal is at the master side. the cclk is used for bit timing of cdin and cdout. the rate of the cclk is config- ured via the cfg command and can be twice of the data rate or at the data rate. the source of that signal is at the master side.
1?ardware ISD-T360SB 1-15 isd figure 1-12: codec interface with one single codec, nsc tp3054, for single line operation table 1-8: typical codec applications application codec type no. of channels master/ slave channel width (no. bits) long/ short frame protocol bit rate cclk freq. (mhz) sample rate (hz) no. of frame syncs analog ? law single 1 master 8 short or long 1 2.048 8000 1 isdn? bit digital?- law dual 2 slave 8 short 1 or 2 2.048 8000 1 linear single 1 master 16 short 1 2.048 8000 1 iom-2/gci single or dual 1? slave 8 short 1 or 2 1.536 8000 1 266 compatibility single or dual 1 or 2 master 8 long or short 1 2.048 8000 1 or 2
1?ardware ISD-T360SB 1-16 voice solutions in silicon figure 1-13: codec interface diagram with two, single codecs, nsc tp3054, and nsc tp3057, for speakerphone operation figure 1-14: codec interface for dual line or single line and speakerphone operation with oki dual codec
1?ardware ISD-T360SB 1-17 isd figure 1-15: codec interface for dual line or single line and speakerphone with lucent dual codec figure 1-16: codec interface for dual line or single line and speakerphone operation with macronix dual codec
1?ardware ISD-T360SB 1-18 voice solutions in silicon 1.3 specifications 1.3.1 absolute maximum ratings note absolute maximum ratings indicate limits beyond which permanent damage may occur. continuous operation at these limits is not intended; operation should be limited to the conditions specified below. 1.3.2 electrical characteristics t a = 0? to +70?, v cc = 5 v ?0% or 3.3 v ?0%, gnd = 0 v storage temperature ?5?c to +150?c temperature under bias 0?c to +70?c all input and output voltages with respect to gnd ?.5 v to +6.5 v table 1-9: electrical characteristics?reliminary information (all parameters with reference to v cc = 3.3 v) symbol parameter conditions min typ max units c x x1 and x2 capacitance 1 17.0 pf i cc1 active supply current normal operation mode, running speech applications 2 40.0 80.0 ma i cc2 standby supply current normal operation mode, dspm idle 2 30.0 ma i cc3 power-down mode supply current power-down mode 2,3 0.7 ma i l input load current 0 v v in v cc ?.0 5.0 m a i o (off) output leakage current (i/o pins in input mode) 0 v v out v cc ?.0 5.0 m a t casa cas active after r.e. cttl, t1 or t2w3 12.0 t cash cas hold after r.e. cttl 0.0 t casia cas inactive after r.e. cttl, t3 or terf 12.0 t caslw dram, pdm, cas width at 0.8 v, both edges 600.0 t dwea dwe active after r.e. cttl, t2w2 12.0 t dweh dwe hold after r.e. cttl 12.0 t dweia dwe inactive after r.e. cttl, t3 12.0 t rasa ras active after r.e. cttl, t2w1 or t2wrf 12.0 t rash ras hold after r.e. cttl 0.0 t rasia ras inactive after r.e. cttl, t3 or t3rf 12.0 t raslw dram pdm, ras width at 0.8 v, both edges 200.0 t rlcl dram pdm ras low, after cas low f.e. cas to f.e. ras 200.0 t wra wr0 active after r.e. cttl, t1 t ctp /2+2
1?ardware ISD-T360SB 1-19 isd 1. guaranteed by design. 2. i out =0, t a 25?c, v cc = 3.3 v for v cc pins and 3.3 v or 5 v on v cchi pins, operating from a 4.096 mhz crystal and running from internal memory with expansion memory disabled. 3. all input signals are tied to 0 (above v cc ?0.5 v or below v ss + 0.5 v), except env0, which is tied to v cc . 4. measured in power-down mode. the total current driven, or sourced, by all the voicedsp processor? output signals is less than 50 ?. 5. guaranteed by design, but not fully tested. t wrcsh wr0 hold after emcs 4 r.e. emcs r.e. to r.e. wr0 10.0 t wrh wr0 hold after r.e. cttl t ctp /2? t wria wr0 inactive after r.e. cttl, t3 t ctp /2+2 v envh env0 input, high voltage 2.0 v v hh cmos input with hysteresis, logical 1 input voltage 2.1 v v hl cmos input with hysteresis, logical 0 input voltage 0.8 v v hys hysteresis loop width 1 0.5 v v ih ttl input, logical 1 input voltage 2.0 v cc + 0.5 v v il ttl input, logical 0 input voltage ?.5 0.8 v v oh logical 1 ttl, output voltage i oh = ?.4 ma 2.4 v v ohwc mmclk, mmdout and emcs logical 1, output voltage i oh = ?.4 ma 2.4 v i oh = ?0 ? 5 v cc ?.2 v v ol logical 0, ttl output voltage i ol = 4 ma 0.45 v i ol = 50 ? 5 0.2 v v olwc mmclk, mmdout and emcs logical 0, output voltage i ol = 4 ma 0.45 v i ol = 50 ? 5 0.2 v v xh clkin input, high voltage external clock 2.0 v v xl clkin input, low voltage external clock 0.8 v table 1-9: electrical characteristics?reliminary information (all parameters with reference to v cc = 3.3 v) symbol parameter conditions min typ max units
1?ardware ISD-T360SB 1-20 voice solutions in silicon 1.3.3 switching characteristics?reliminary de?itions all timing specifications in this section refer to 0.8 v or 2.0 v on the rising or falling edges of the signals, as illustrated in figure 1-17 through figure 1-23, unless specifically stated otherwise. maximum times assume capacitive loading of 50pf. clkin crystal frequency is 4.096 mhz. note cttl is an internal signal and is used as a ref- erence to explain the timing of other signals. see figure 1-37. figure 1-17: synchronous output signals (valid, active and inactive) note: signal valid, active or inactive time, after a rising edge of cttl or mwclk. figure 1-18: synchronous output signals (valid) note: signal valid time, after a falling edge of mwclk. 2.0v 0.8v cttl or mwclk signal t signal 2.0v 2.0v 0.8v mwclk signal t signal 0.8v
1?ardware ISD-T360SB 1-21 isd figure 1-19: synchronous output signals (hold), after rising edge of cttl note: signal hold time, after a rising edge of cttl. figure 1-20: synchronous output signals (hold), after falling edge of mwclk note: signal hold time, after a falling edge of mwclk. figure 1-21: synchronous input signals note: signal setup time, before a rising edge of cttl or mwck, and signal hold time after a rising edge of cttl or mwck 2.0v 0.8v cttl signal 2.0v t signal 2.0v 0.8v mwclk signal t signal 2.0v 2.0 v cttl or mwclk 2.0 v 0.8 v 2.0 v 0.8 v t signal setup t signal hold signal
1?ardware ISD-T360SB 1-22 voice solutions in silicon figure 1-22: asynchronous signals note: signal b starts after rising or falling edge of signal a. the reset has a schmitt trigger input buffer. figure 1-23 shows the input buffer characteristics. figure 1-23: hysteresis input characteristics 2.0 v 0.8 v 2.0 v 0.8 v signal a signal b t signal v out v hh v hl v hys
1?ardware ISD-T360SB 1-23 isd 1.3.4 synchronous timing tables in this section, r.e. means rising edge and f.e. means falling edge. 1. in normal operation mode, t ctp must be 48.8 ns; in power-down mode, t ctp must be 50,000 ns. 2. guaranteed by design, but not fully tested. table 1-10: output signals?reliminary symbol figure description reference conditions min (ns) max (ns) t ah address hold after r.e. cttl 0.0 t av address valid after r.e. cttl, t1 12.0 t cclka cclk active after r.e. cttl 12.0 t cclkh cclk hold after r.e. cttl 0.0 t cclkia cclk inactive after r.e. cttl 12.0 t cdoh cdout hold after r.e. cttl 0.0 t cdov cdout valid after r.e. cttl 12.0 t ctp cttl clock period 1 r.e. cttl to next r.e. cttl 30.5 250,000 t emcsa emcs active after r.e. cttl, t2w1 12.0 t emcsh emcs hold after r.e. cttl 0.0 t emcsia emcs inactive after r.e. cttl t3 12.0 t fsa cfs0 active after r.e. cttl 25.0 t fsh cfs0 hold after r.e. cttl 0.0 t fsia cfs0 inactive after r.e. cttl 25.0 t mmclka master microwire clock active after r.e. cttl 12.0 t mmclkh master microwire clock hold after r.e. cttl 0.0 t mmclkia master microwire clock inactive after r.e. cttl 12.0 t mmdoh master microwire data out hold after r.e. cttl 0.0 t mmdov master microwire data out valid after r.e. cttl 12.0 t mwdof microwire data float 1 after r.e. mwcs 70.0 t mwdoh microwire data out hold 2 after f.e. mwclk 0.0 t mwdonf microwire data no float 2 after f.e. mwcs 0.0 70.0 t mwdov microwire data out valid 2 after f.e. mwclk 70.0 t mwitop mwdin to mwdout propagation time 70.0 t mwrdya mwrdy active after r.e. of cttl 0.0 35.0 t mwrdyia mwrdy inactive after f.e. mwclk 0.0 70.0 t pabch pb and mwrqst after r.e. cttl 0.0 t pabcv pb and mwrqst after r.e. cttl, t2w1 12.0
1?ardware ISD-T360SB 1-24 voice solutions in silicon 1. guaranteed by design, but not fully tested in power-down mode. 2. guaranteed by design, but not fully tested. table 1-11: input signals?reliminary symbol figure description reference conditions min (ns) max (ns) t cclksp codec clock period (slave) r.e. cclk to next r.e. cclk 244 t cclksh codec clock high (slave) at 2.0 v (both edges) 120 t cclksl codec clock low (slave) at 0.8 v (both edges) 120 t cdih cdin hold after r.e. cttl 0.0 t cdis cdin setup before r.e. cttl 11.0 t cfs0ss cfs0 setup before r.e. cclk tbd t cfs0sh cfs0 hold after r.e. cclk tbd t dih data in hold (d0:7) after r.e. cttl t1, t3 or ti 0.0 t dis data in setup (d0:7) before r.e. cttl t1, t3 or ti 15.0 t mmdinh master microwire data in hold after r.e. cttl 0.0 t mmdins master microwire data in setup before r.e. cttl 11.0 t mwckh microwire clock high (slave) at 2.0 v (both edges) 100.0 t mwckl microwire clock low (slave) at 0.8 v (both edges) 100.0 t mwckp microwire clock period (slave) 1 r.e. mwclk to next r.e. mwclk 2.5 ? t mwclkh mwclk hold after mwcs becomes inactive 50.0 t mwclks mwclk setup before mwcs becomes active 100.0 t mwcsh mwcs hold after f.e. mwclk 50.0 t mwcss mwcs setup before r.e. mwclk 100.0 t mwdih mwdin hold after r.e. mwclk 50.0 t mwdis mwdin setup before r.e. mwclk 100.0 t pwr power stable to reset r.e. 2 after v cc reaches 4.5 v 30.0 ms t rstw reset pulse width at 0.8 v (both edges) 10.0 ms t xh clkin high at 2.0 v (both edges) t x1p /2 ?5 t xl clkin low at 0.8 v (both edges) t x1p /2 ?5 t xp clkin clock period r.e. clkin to next r.e. clkin 244.4
1?ardware ISD-T360SB 1-25 isd 1.3.5 timing diagrams figure 1-24: rom read cycle timing 1. this cycle may be either ti (idle), t3 or t3h. 2. data can be driven by an external device at t2w1, t2w, t2 and t3. 3. this cycle may be either ti (idle) or t1. figure 1-25: aram/dram refresh cycle timing (normal operation) 1. this cycle may be either ti (idle) or t1 of any non-dram bus cycle. if the next bus cycle is a dram one, t3rf is followed by three ti (idle) cycles.
1?ardware ISD-T360SB 1-26 voice solutions in silicon figure 1-26: aram/dram power-down refresh cycle timing figure 1-27: dram read cycle timing 1. a0-a10 in the ire environment, when expansion memory is disabled; otherwise a0-a15. 2. this cycle may be either ti (idle) or t1 of any non-dram bus cycle. if the next bus cycle is to dram, t3 is followed by three ti (idle) cycles. 3. an external device can drive data from t2w3 to t3. cttl ras c as a0-10 d we d0-1, d3-7 d2/ra11 ti ti t1 t2w1 t2w1 t2w2 6xt2w ti t1 t2 t3 ti row column t rasa t rasia t casa t casia t av d2 in t df t df t av t dis t dih t dis t dih t df (note 1) (note 3) (note 3) t2w2 t2w3 (note 2) data in ra11 t ah t av t ah t rash t rash t cash t cash t2w3 ra11 a0-15 or
1?ardware ISD-T360SB 1-27 isd figure 1-28: dram write cycle timing 1. a0?10 in the ire environment, when expansion memory is disabled; otherwise a0?15. 2. this cycle may be either ti (idle) or t1 of any non-dram bus cycle. if the next bus cycle is to dram, t3 is followed by three ti (idle) cycles. cttl ras c as d we d0-1, d3-7 d2/ra11 ti ti t1 t2w1 t2w2 6xt2w ti t2 t3 t rasa t rasia t casa t casia t df (note 1) (note 2) t dwea t df t dh t dv t df t dh row column d2 out data out ra11 t av t ah t rash t rash t cash t cash t dweh t dweia t dweh t av t av t dv t2w3 a0-10 a0-15 or
1?ardware ISD-T360SB 1-28 voice solutions in silicon figure 1-29: codec short frame timing note: the cclk and cfs0 timing is shown for master mode only. for slave mode, see figure 1-31. figure 1-30: codec long frame timing note: the cclk and cfs0 timing is shown for master mode only. for slave mode, see figure 1-31. cttl cclk cfs0/ cdout cdin t ctp t ctp t ctp t ctp t ctp t ctp 4xt ctp t ctp t ctp t ctp t ctp 4xt ctp t cclka t cclkia t fsa t ctp t ctp cfs1 t fsia t cdis t cdih bit 7 bit 7 t cdov t cdoh t cclkh t cclkh t fsh t fsh t ctp t ctp 4xt ctp t ctp t ctp t ctp t ctp 4xt ctp t ctp t ctp t cdov cttl cclk cfs0/ cdout cdin t ctp t ctp t ctp t ctp t ctp t ctp 4xt ctp t ctp t ctp t ctp t ctp 4xt ctp t cclka t cclkia t fsa t ctp cfs1 t fsia bit 0 t cdov t cdoh t cclkh t cclkh t fsh t fsh t ctp t ctp 4xt ctp t ctp t ctp t ctp bit 7 bit 7
1?ardware ISD-T360SB 1-29 isd figure 1-31: slave codec cclk and cfs0 timing figure 1-32: microwire transaction timing?ata transmitted to output cclk cfs0 t cfs0ss t cfs0sh t cclksh t cclksl t cclksp
1?ardware ISD-T360SB 1-30 voice solutions in silicon figure 1-33: microwire transaction timing?ata echoed to output figure 1-34: master microwire timing
1?ardware ISD-T360SB 1-31 isd figure 1-35: output signal timing for port pb and mwrqst note: this cycle may be either ti (idle), t2, t3 or t3h. figure 1-36: clkin timing figure 1-37: cttl timing
1?ardware ISD-T360SB 1-32 voice solutions in silicon figure 1-38: reset timing when reset is not at power-up figure 1-39: reset timing when reset is at power-up
2?oftware ISD-T360SB 2-1 isd chapter 2?software the voicedsp software resides in the on-chip rom. it includes dsp-based algorithms, system support functions and a software interface to hardware peripherals. 2.1 system operation this section provides details of the system sup- port functions and their principle operation. it is divided into the following subjects: the state machine command execution event handling message handling tone generation initialization and configuration power down mode (pdm) 2.1.1 the state machine the ISD-T360SB operates in two modes, normal mode (dtad) and speakerphone mode. to change the mode use the set speakerphone mode (ssm) command. the voicedsp processor functions as a state machine under each mode. it changes state either in response to a com- mand sent by the microcontroller, after execu- tion of the command is completed, or as a result of an internal event (e.g. memory full or power failure). for more information see ?oicedsp processor commands?uick reference table?on page 2-21. the voicedsp states are desctibed below: reset the voicedsp processor is initialized to the reset state after a full hardware reset by the reset sig- nal (see ?esetting?on page 1-3). in this state the processor detectors (vox, constant energy, call progress tones and dtmf) are not active. in all other states, these detectors are active. (see the sdet and rdet commands for further de- tails). idle this is the state from which most commands are executed. as soon as a command and all its pa- rameters are received, the processor starts exe- cuting the command. play in this state a message is decompressed (unless stored in pcm format), and played back. record in this state a message is compressed (unless stored in pcm format) and recorded into the message memory. synthesis an individual word or a sentence is synthesized from an external vocabulary in this state. tone_generate in the tone_generate state, the voicedsp pro- cessor generates single or dtmf tones. msg_open the voicedsp processor either reads or writes 32 bytes to the message memory, or sets the mes- sage read/write pointer on a 32 byte boundary.
2?oftware ISD-T360SB 2-2 voice solutions in silicon 2.1.2 command execution a voicedsp processor command is represented by an 8-bit opcode. some commands have pa- rameters and some have return values. com- mands are either synchronous or asynchronous. synchronous commands a synchronous command must complete exe- cution before the microcontroller can send a new command (e.g. gms, gew). a command sequence begins when the microcontroller sends an 8-bit opcode to the processor, fol- lowed by the command? parameters (if any). the voicedsp processor then executes the com- mand and, if required, transmits a return value to the microcontroller. upon completion, the pro- cessor notifies the microcontroller that it is ready to accept a new command. asynchronous commands an asynchronous command starts execution in the background and notifies the microcontroller, which can send more commands while the cur- rent command is still running (e.g. r, p). after re- ceiving an asynchronous command, such as p (playback), r (record), sw (say words) or gt (generate tone), the voicedsp processor switches to the appropriate state and executes the command until finished or a s (stop) or pa (pause) command is received from the micro- controller. when completed, the ev_normal _end event is set and the processor switches to the idle state. ?oicedsp processor commands?uick reference table?on page 2?1 displays all the processor commands, the valid source states in which these commands are valid, and the states resulting from the command. 2.1.3 event handling status word the 16-bit status word indicates events that oc- cur during normal operation. the voicedsp pro- cessor activates the mwrqst signal, to indicate a change in the status word. this signal remains active until the processor receives a gsw (get status word) command. for detailed description of the status word and the meaning of each bit, see ?sw get status word?on page 2-33. error word the 16-bit error word indicates errors that oc- curred during execution of the last command. if an error is detected, the command is not pro- cessed; the ev_error bit in the status word is set to 1, and the mwrqst signal is activated. error handling when the microcontroller detects the active mwrqst signal, it issues the gsw command, de- activating the mwrqst signal. then, the micro- controller tests the ev_error bit in the status word, and, if set, sends the gew (get error word) command to read the error word for details. for detailed description of the error word and the meaning of each bit, see ?ew get error word?on page 2-30.
2?oftware ISD-T360SB 2-3 isd 2.1.4 message handling a message is the basic unit on which most of the voicedsp commands operate. a voicedsp pro- cessor message, stored on a memory device (flash or aram/dram), can be regarded as a computer file stored on a mass-storage device. the ISD-T360SB manages messages for a wide range of applications, which require different levels of dtad functionality. the voicedsp pro- cessor features advanced memory-organization features such as multiple outgoing messages (ogms), mailboxes, and the ability to distinguish between incoming messages (icms) and ogms. a message is created with either the r (record) or the cmsg (create message) command. once created, the message is assigned a time- and-day stamp and a message tag which is read by the microcontroller. the r command takes voice samples from the codec, compress- es them, and stores them in the message mem- ory. when a message is created with the cmsg command the data to be recorded is provided by the microcontroller, via the wmsg (write mes- sage) command and not through the codec. here, the data is transferred directly to the mes- sage memory, and not compressed by the isd- t360sb voice compression algorithm. wmsg, rmsg (read message) and smsg (set message pointer) are message-data access commands used to store and read data to or from any location in the message memory (see ?oicedsp processor commands?uick reference table?on page 2-21 for more de- tails). using these commands, the microcontrol- ler utilizes messages for features such as a telephone directory and playing back (p com- mand) or deleting (dm command) a message. remove redundant data (e.g., trailing tones or silence) from the message tail with the cmt (cut message tail) command. the pa (pause) and res (resume) commands suspend the p and r commands, respectively, and then resume them from where they were suspended. current message the gtm (get tagged message) command se- lects the current message. most message han- dling commands (p, dm, rmsg), operate on the current message. deleting the prevailing message does not cause a different message to become current; the cur- rent message is undefined. if you issue the gtm command to skip to the next message, the first message, newer than the just deleted message, becomes the current message. message tag each message has a 2-byte message tag which used to categorize messages, and implement such features as outgoing messages, mailbox- es, and different handling of old and new mes- sages. the tag is created during the r (record) com- mand. use the gmt (get message tag) and smt (set message tag) commands to handle mes- sage tags. note message tag bits can only be cleared and are set only when a message is first cre- ated. this limitation, inherent in flash memories, allows bits to be changed only from 1 to 0 (changing bits from 0 to 1 requires a special erasure procedure). however, the usual reason for updating an existing tag is to mark a message as old. this can be done when a message is first cre- ated by using one of the bits as a new/old indicator, setting the bit to 1 and later clearing it when necessary.
2?oftware ISD-T360SB 2-4 voice solutions in silicon 2.1.5 tone generation the voicedsp processor generates dtmf tones and single-frequency tones from 300hz to 3000hz in increments of 100hz. the ISD-T360SB tone gen- eration conforms to the eia-470-rs standard. note, however, that value of some tunable pa- rameters may need adjusting to meet the stan- dard specifications since the energy level of generated tones depends on the analog circuits used. 1. tune the dtmf_gen_twist_level param- eter to control the twist level of the gener- ated dtmf tones. 2. use the vc (volume control) command, and tune the tone_gen_level parame- ter, to control the energy level at which these tones are generated. 3. use the gt (generate tone) command to specify the dtmf tones, and the frequen- cy at which single tones are generated. refer to table 2-5, vc command and gt com- mand of the command description section for furhter details of the relevant tunable parame- ters and commands. note the dtmf detector performance is degraded during tone generation, espe- cially if the frequency of the generated tone is close to the frequency of one of the dtmf tones. 2.1.6 initialization and configuration use the following procedures to initialize the voicedsp processor: normal initialization reset the voicedsp processor by activating the reset signal. (see ?esetting?on page 1-3.) 1. issue a cfg (configure voicedsp proces- sor) command to change the configura- tion according to your environment. 2. issue an init (initialize system) command to initialize the voicedsp firmware. 3. issue a series of tune commands to adjust the voicedsp processor to the require- ments of your system. tunable parameters the voicedsp processor can be adjusted to the specific system? requirements using a set of tun- able parameters. these parameters are set to their default values after reset and can be later modified with the tune command. by tuning these parameters, you can control various as- pects of the voicedsp processor? operation, such as silence compression, tone detection, and no-energy detection. tables 2-4 to 2-11 of the command description section describe all the tunable parameters in detail.
2?oftware ISD-T360SB 2-5 isd 2.1.7 power-down mode the pdm (go to power-down mode) command switches the ISD-T360SB to power-down mode. the purpose of the pdm command is to save power during buttery operation, or for any other power saving cause. during power-down mode only basic functions, such as aram/dram re- fresh and time and date update, are active (for more details refer to power-down mode de- scription on page 1-4). this pdm command may only be issued when the processor is in the idle mode (for an expla- nation of the ISD-T360SB states, see ?ommand execution?on page 2?). if it is necessary to switch to power-down mode from any other state, the controller must first issue a s (stop) command to switch the processor to the idle state, and then issue the pdm command. send- ing any command while in power-down mode resets the voicedsp processor detectors, and re- turns it to normal operation mode. note entering or exiting power-down mode can distort the real-time clock by up to 500 ?. thus, to maintain the accuracy of the real- time clock, enter or exit the power-down mode as infrequently as possible. 2.2 peripherals this section provides details of the peripherals in- terface support functions and their principle op- eration. it is divided into the following subjects: microcontroller interface (slave microwire) memory interface codec interface 2.2.1 microcontroller interface microwire/plus is a synchronous serial com- munication protocol minimizes the number of connections, and thus the cost, of communicat- ing with peripherals. the voicedsp microwire interface implements the microwire/plus interface in slave mode, with an additional ready signal. it enables a mi- crocontroller to interface efficiently with the voicedsp processor application. the microcontroller is the protocol master and provides the clock for the protocol. the voiced- sp processor supports clock rates of up to 400 khz. this transfer rate refers to the bit transfer; the actual throughput is slower due to byte process- ing by the voicedsp processor and the micro- controller. communication is handled in bursts of eight bits (one byte). in each burst the voicedsp proces- sor is able to receive and transmit eight bits of data. after eight bits have been transferred, an internal interrupt is issued for the voicedsp pro- cessor to process the byte, or to prepare another byte for sending. in parallel, the voicedsp pro- cessor sets mwrdy to 1, to signal the microcon- troller that it is busy with the byte processing. another byte can be transferred only when the mwrdy signal is cleared to 0 by the voicedsp processor. when the voicedsp processor trans- mits data, it expects to receive the value 0xaa before each transmitted byte. the voicedsp processor reports any status change by clearing the mwrqst signal to 0. if processor command? parameter is larger than one byte, the microcontroller transmits the most significant byte (msb) first. if a return value is larg- er than one byte, the voicedsp processor trans- mits the msb first. the following signals are used for the interface protocol. input and output are relative to the voicedsp processor.
2?oftware ISD-T360SB 2-6 voice solutions in silicon input signals mwdin microwire data in. used for input only, for transferring data from the microcontroller to the voicedsp processor. mwclk microwire clock. serves as the synchronization clock during communication. one bit of data is transferred on every clock cycle. the input data is available on mwdin and is latched on the clock rising edge. the transmitted data is output on mwdout on the clock falling edge. the sig- nal should remain low when switching mwcs . mwcs microwire chip select. the mwcs signal is cleared to 0, to indicate that the voicedsp pro- cessor is being accessed. setting mwcs to 1 causes the voicedsp processor to start driving mwdout with bit 7 of the transmitted value. set- ting the mwcs signal resets the transfer-bit counter of the protocol, so the signal can be used to synchronize between the voicedsp pro- cessor and the microcontroller. to prevent false detection of access to the voicedsp processor due to spikes on the mwclk signal, use this chip select signal, and toggle the mwclk input signal, only when the voicedsp processor is accessed. output signals mwdout microwire data out. used for output only, for transferring data from the voicedsp processor to the microcontroller. when the voicedsp proces- sor receives data it is echoed back to the micro- controller on this signal, unless the received data is 0xaa. in this case, the voicedsp processor ech- oes a command? return value. mwrd y microwire ready. when active (0), this signal indicates that the voicedsp processor is ready to transfer (receive or transmit) another byte of da- ta. this signal is set to 1 by the voicedsp processor after each byte transfer has been completed. it remains 1, while the voicedsp processor is busy reading the byte, writing the next byte, or exe- cuting the received command (after the last pa- rameter has been received). mwrdy is cleared to 0 after reset. for proper operation after a hardware reset, this signal should be pulled up. mwrqst microwire request. when active (0), this signal indicates that new status information is avail- able. mwrqst is deactivated (set to 1), after the voicedsp processor receives a gsw (get status word) command from the microcontroller. after reset, this signal is active (0) to indicate that a re- set occurred. mwrqst , unlike all the signals of the communication protocol, is an asynchro- nous line that is controlled by the voicedsp firm- ware. signal use in the interface protocol after reset, both mwrqst and mwrdy are cleared to 0. the mwrqst signal is activated to indicate that a reset occurred. the ev_reset bit in the status register is used to indicate a reset condition. the gsw command should be issued after reset to verify that the ev_reset event occurred, and to deactivate the mwrqst signal. while the mwcs signal is active (0), the voiced- sp processor reads data from mwdin on every rising edge of mwclk. voicedsp processor also writes every bit back to mwdout. this bit is either the same bit which was read from mwdin (in this case it is written back as a synchronization echo after some propagation delay), or it is a bit of a value the voicedsp processor transmits to the microcontroller (in this case it is written on every falling edge of the clock).
2?oftware ISD-T360SB 2-7 isd when a command has more than one parame- ter/return-value, the parameters/return-values are transmitted in the order of appearance. if a parameter/return-value is more than one byte long, the bytes are transmitted from the most sig- nificant to the least significant. the mwrdy signal is used as follows: 1. active (0) mwrdy signals the microcon- troller that the last eight bits of data trans- ferred to/from the voice module were accepted and processed (see below). 2. the mwrdy signal is deactivated (set to 1 by the voicedsp processor) after 8-bits of data were transferred to/from the voicedsp processor. the bit is set follow- ing the falling edge of the eighth mwclk clock-cycle. 3. the mwrdy signal is activated (cleared to 0) by the voicedsp processor when it is ready to receive the first parameter byte (if there are any parameters) and so on till the last byte of parameters is transferred. an active mwrdy signal after the last byte of parameters indicates that the command was parsed and (if possible) executed. if that command has a return value, the microcontroller must read the value before issuing a new command. 4. when a return value is transmitted, the mwrdy signal is deactivated after every byte, and activated again when the voicedsp processor is ready to send an- other byte, or to receive a new com- mand. 5. the mwrdy signal is activated (cleared to 0) after reset, and after a protocol time- out. (see ?nterface protocol time- outs?) the mwrqst signal is used as follows: 1. the mwrqst signal is activated (cleared to 0), when the status word is changed. 2. the mwrqst signal remains active (0), un- til the voicedsp processor receives a gsw command. figure 1-32 and figure 1-33 illustrate the se- quence of activities during a microwire data transfer between voicedsp and the microcon- troller. interface protocol time-outs depending on the voicedsp processor? state, if more than 100 milliseconds elapse between the assertion of the mwrdy signal and the transmis- sion 8th bit of the next byte pertaining to the same command transaction, a time-out event occurs, and the voicedsp processor responds as follows: 1. sets the error bit in the status word to 1. 2. sets the ev_timeout bit in the error word to 1. 3. activates the mwrqst signal (clears it to 0). 4. activates the mwrdy signal (clears it to 0). 5. waits for a new command. (after a time- out occurs, i.e., the microcontroller re- ceived mwrqst during the command transfer, or result reception, the microcon- troller must wait at least four milliseconds before issuing the next command.) echo mechanism the voicedsp processor echoes back to the mi- crocontroller all the bits received by the voiced- sp processor. upon detection of an error in the echo, the microcontroller should stop the proto- col clock, which eventually causes a time-out er- ror (i.e., err_timeout bit is set in the error word). note when a command has a return value, the voicedsp processor transmits bytes of the return value instead of the echo value. the voicedsp processor transmits a byte as an echo when it receives the value 0xaa from the microprocessor. upon detection of an error the voicedsp processor activates the mwrqst sig- nal, and sets the err_comm bit in the error word.
2?oftware ISD-T360SB 2-8 voice solutions in silicon 2.2.2 memory interface device number and type the voicedsp processor supports various types of flash memory and aram/dram devices. up to four devices may be connected to the voicedsp, where all the connected devices must be of the same type. each memory device may be of 4mbit, 8mbit or 16mbit; thus a total of 64mbit non-volatile memory may be connected for message storage (up to 4 hours of voice re- cording). see ?emory interface?on page 1-6, for de- tailed description of the supported flash and aram/dram devices and the hardware con- nectivity. use the cfg command to define the type and number of installed memory devices (see ?fg configure voicedsp config_value?on page 2- 25). memory device size the memory manager handles the memory de- vices in basic units of 4kbyts blocks. this ap- proach is defined due to the nature of flash devices where the basic unit that can be erased is a 4kbytes block. this constraint is not relevant for aram/dram devices, but the concept is maintained for simplicity and consistency. memory blocks cannot be shared by different voice messages. therefore, the maximum num- ber of messages per memory device, equals to the number of memory blocks minus one (one block per device is used for memory manage- ment). the size of the connected memory devices, is defined by the number of memory blocks in each device. refer to tunable parameter index 62, in table 2-10, for detailed description of the available number of blocks for flash and aram/ dram devices. production line testing in many cases it is desired to test the ISD-T360SB in the production line as part of the whole appli- cation. usually in these cases, the testing time is an important factor and should be minimized as possible. the initialization time of the memory de- vices is significant and should be avoided during production (refer to table 1-4). therefore, a dedicated parameter is defined in order to allow a production line testing while using a small part of the real connected memory size. it should be noted that in case of power failure during the production line testing, the connect- ed memory devices should be replaced, and the process should be repeated. refer to pa- rameter index 63, in table 2-10, for further expla- nation of the production line testing. aram quality aram is an audio grade ram device, which im- plies that some percentage of the aram bits are defected and their content is undefined. unlike flash devices, where the defected bits can be mapped out, in case of aram specific bits can- not be mapped out; only memory blocks can be mapped out. therefore, it should be noted that using aram as a voice storage device, will result in audible dis- tortions. it is the user responsibility to define the maximum allowed bad nibbles (4 bits) in a mem- ory block. if the number of bad nibbles exceeds the defined limitation, the specific block is mapped out and is not used for message re- cording. refer to tunable parameter index 64, in table 2-10, for further details of the aram quality level definition.
2?oftware ISD-T360SB 2-9 isd 2.2.3 codec interface supported functionality the voicedsp processor supports analog and digital telephony in various configurations. for analog telephony the voicedsp operates in master mode, where it provides the clock and the synchronization signals. it supports a list of sin- gle channel and dual channel codecs, as listed in table 1-7. for digital telephony the voicedsp operates in slave mode, where the control sig- nals are provided by an external source. the codec interface is designed to exchange data in short frame format as well as in long frame format. the channel width may be either 8 bits (u-law format or a-law format), or 16 bits (linear format). in slave mode the clock may be divided by two, if required (two bit rate clock mode). the voicedsp support up to 2 voice channels, where the line should be connected as channel 0 (in master mode or in slave mode - depends on the configuration), and the speakerphone (speaker and microphone) should be connect- ed as channel 1 or as channel 2, depends on the configuration (channel 1 and channel 2 are al- ways connected as master). see ?he codec interface?on page 1-13, for de- tailed description of the supported codec de- vices and the hardware connectivity. use the cfg command to define the codec mode (master or slave), the data frame format (short or long), the channel width (8 bits or 16 bits), the clock bit rate (single or dual) and the number and type of codecs (one or two, single channel or dual channel). see ?fg configure voicedsp config_value?on page 2-25. data channels timing especially in digital telephony, but also in analog telephony when speakerphone is connected, the channels data may be delayed from the synchronization signal by variable number of clock cycles. in order to allow full flexibility of the data delay relative to the synchronization signal, and the delay between the two synchronization signals, a set of registers is provided. setting the delay parameters of these registers defines the exact timing of all the codec interface signals. refer to tunable parameters index 65 to index 69, in table 2-11, for detailed description of the delay registers and their significance. 2.3 algorithm features this section provides details of the voicedsp al- gorithms and their principle operation. it is divid- ed into the following subjects: vcd (voice compression and decompression) dtmf detection tone and energy detection (call progress) speakerphone speech synthesis
2?oftware ISD-T360SB 2-10 voice solutions in silicon 2.3.1 vcd (voice compression and decompression) the voicedsp processor implements a state of the art vcd algorithm of the celp family. the al- gorithm provides 3 compression rates that can be selected dynamically (actually, the algorithm supports more compression rates). pcm record- ing (no compression) is also provided. the lowest compression rate of 5.3 kbit/s enables about 30 minutes of recording on an 8-mbit de- vice (depending on the relative silence period). the mid-quality compression rate of 9.9 kbit/s provides about 16 minutes of voice recording time. the highest compression rate of 16.8 kbit/s, the highest quality recording, stores up to 10 min- utes on a 8-mbit device. for detailed information about recording times refer to table 1-5. before recording each message, the microcon- troller selects one of the three compression rates, or pcm recording, with the compression_rate parameter of the r (record) command. during message playback the voicedsp processor reads this one byte parameter and selects the appropriate speech decompression algorithm. ivs vocabularies can be prepared in either of the three compression rates, or in pcm format, using the ivs tool. all messages in a single vocab- ulary must be recorded using the same algo- rithm. (see the ivs user? guide for more details). during speech synthesis, the voicedsp processor automatically selects the appropriate speech decompression algorithm. silence compression a voice activity detector (vad) is used in order to detect periods of silence during the compres- sion of the recorded message. silence is treated differently than normal voice by the compres- sion algorithm. it is compressed to about 1.0 kbit/ s. the compressed silence contains data that al- lows to generate comfort noise during message playback. the comfort noise generation is impor- tant because the human ear is not used to ?eal silence while listening to messages. various tunable parameters are available in or- der to optimally tune the vad. the silence com- pression may be turned off, though it is planned to remain on continuously. for more details refer to table 2-4 of the command description sec- tion. note the silence compression should be turned off when aram devices are used for voice storage. otherwise, unpredictable results are expected during message playback. sw agc a software automatic gain control (sw agc) algorithm is activated with the compression module in order to regulate the input signal to a dynamic range that will provide higher compres- sion quality. the algorithm senses the energy lev- el and updates the signal gain in order to amplify low energy signals and to avoid signal satura- tion. the sw agc feature eliminates the need for an external hw agc, thus reducing hardware costs and complexity. hardware gain control may be used to avoid signal saturation prior to sampling the signal. a tunable parameter determines the maximum allowed gain for the sw agc algorithm. the sw agc may be turned off, though it is planned to remain on continuously. for more details refer to table 2-4 of the command description section. variable speed playback this feature increases or decreases the speed of messages and synthesized messages during playback. use the sps (set playback speed) to set the speed of message playback. the new speed applies to all recorded messages and syn- thesized messages (only if synthesized using ivs), until changed by another sps command. if this command is issued while the voicedsp proces- sor is in the play state, the speed also changes for the message currently being played. the speedup / slowdown algorithm is designed to maintain the pitch of the original speech. this approach provides the same speech tone while playback speed varies.
2?oftware ISD-T360SB 2-11 isd pcm recording the voicedsp is capable of recording data in pcm format (that is the original samples format either in 8 bits u-law format, 8 bit a-law format or 16 bits linear format). the pcm data uses more storage space, but it provides the highest quality for ogm, music-on-hold or ivs data. the pcm re- cording may be selected as one of the available compression rates during the r command (compression_rate = 0). silence compression and variable speed play- back are not feasible during pcm recording and playback since this feature skips the compres- sion algorithm. 2.3.2 dtmf detection the voicedsp processor detects dtmf, which enables remote control operations. detection is active throughout the operation of the voiced- sp processor. detection can be configured using the sdet (set detectors mask) command, which controls the reporting of the occurrence of tones, and the rdet (reset detectors) com- mand which resets the detectors. the accuracy of the tone length, as reported by the tone de- tectors, is ?0 ms. dtmf detection may be reported at the starting point, ending point, or both. the report is made through the status word (for further details, see gsw command). for further details about tunable parameters re- fer to table 2-6 of the command description sec- tion. the dtmf detector performance, as measured on the line input using an isd-ds360-daa board, is summarized below (see table 2-1). 1. performance depends on the daa design. for reliable dtmf detection: - a hardware echo-canceler, that attenuates the echo by at least 6 dbm, is required during playback. - the hw agc, if present, must be disabled during playback. 2. performance with echo canceler is 10 db better than without echo canceler. for a silent message, detection sensitivity is - 34 dbm, with echo canceler. 3. tune parameters 60 and 61 may improve dtmf detection sensitivity. for more details refer to the parameters description in table 2-6. 4. the accuracy of reported dtmf tones is ?0 ms. 5. if the interval between two consecutive identical dtmf tones is less than, or equal to, 20 ms, the two are detected as one long dtmf tone. if the interval between two consecutive identical dtmf tones is between 20 ms and 45 ms, separate detection is unpredictable. 6. determined by the dtmf_rev_twist tunable parameter value. table 2-1: dtmf detector performance 1 play/ivs synthesis record/idle detection sensitivity performance depends on the message being played. 2 - 34 dbm accepted dtmf length 3 >50 ms >40 ms frequency tolerance ?.5% ?.5% s/n ratio 12 db 12 db minimum spacing 4 >50 ms >45 ms normal twist 8 db 8 db reverse twist 5 4 db or 8 db 4 db or 8 db
2?oftware ISD-T360SB 2-12 voice solutions in silicon dtmf sw agc in order to remove the linkage between the hw agc and the detection level of the dtmf detec- tor, two new tunable parameters are added. these tunable parameters define the gain of the sw agc for dtmf signals. dtmf_det_agc_idle - sw agc for dtmf detec- tion in idle and record states. when increment- ing this tunable by 1, the dynamic range is increased by 3 db. dtmf_det_agc_play - sw agc for dtmf detec- tion in play and tone_generate states. when in- crementing this tunable by 1, the dynamic range is increased by 3 db. echo cancellation echo cancellation is a technique used to im- prove the performance of dtmf detection dur- ing speech synthesis, tone generation, and ogm playback. for echo cancellation to work properly, hw agc must not be active in parallel. thus, to take advantage of echo cancellation, the microcontroller must control the hw agc, if exists, (i.e., disable the hw agc during play, synthesis and tone_generate states and en- able it again afterwards). if hw agc can not be disabled, do not use echo cancellation. the microcontroller should use the cfg com- mand to activate/deactivate echo cancella- tion. note normally, a hw agc is not required with the ISD-T360SB, since sw agc is active for the vcd algorithm, dtmf detection and the speakerphone module. 2.3.3 tone and energy detection (call progress) the voicedsp processor detects busy and dial tones, constant energy level, and no-energy (vox). this enables call progress tracking. de- tection is active throughout the operation of the voicedsp processor. detection can be config- ured using the sdet (set detectors mask) com- mand, which controls the reporting of the occurrence of tones, and the rdet (reset de- tectors) command which resets the detectors. the accuracy of the tone length, as reported by the tone detectors, is ?0 ms. tunable parameters tunable parameters control the detection of busy and dial tones, constant energy level (in the frequency range 200?400hz), and no-ener- gy. these parameters should be tuned to fit the system hardware. in addition, changes may be required to the tunable parameters according to the setting (on or off) of the hw automatic gain control (hw agc), if exists. for more infor- mation refer to tables 2-7, 2-8 of the command description section.
2?oftware ISD-T360SB 2-13 isd figure 2-1: busy and dial-tone band-pass filter frequency response busy and dial tones busy and dial-tone detectors work with a band- pass filter that limits the frequency range in which tones can be detected to 0?100hz. figure 1-1 shows the frequency response of this band-pass filter. the design of the busy-tone detector allows very high flexibility in detecting busy tones with vary- ing cadences. the tunable parameters are divided into five sets: 1. busy tone on-time and off-time range specification: busy_det_min_on_time busy_det_min_off_time busy_det_max_on_time busy_det_max_off_time 2. busy tone cadence control specification busy_det_verify_count busy_det_tone_type busy_det_diff_threshold busy_det_verify_count determines the number of on/off cadences that detec- tor should detect before reporting busy tone presence. busy_det_diff_threshold describes the maximum allowed difference between two compared on or off periods, as de- termined by the busy_det_tone_type tunable parameter. busy_det_tone_type specifies the type of cadences that are supported. legal values are: two cadences only three cadences only both two and three cadences. the acceptance criteria for two cad- ences: [e1?3] < busy_det_diff_threshold and [s1?3] < busy_det_diff_threshold the acceptance criteria for three cadences: [e1?4] < busy_det_diff_threshold and [s1?4] < busy_det_diff_threshold 3. busy and dial tone energy thresholds tone_det_on_energy threshold tone_det_off_energy threshold 4. busy detection time busy_det_min_time 0 200 400 600 800 1000 1200 1400 1600 1800 2000 frequency (hz) ?0 ?0 ?0 ?0 0 magnitude db
2?oftware ISD-T360SB 2-14 voice solutions in silicon figure 2-2: busy-tone detector?efault cadence specification constant energy the constant-energy detector reports the pres- ence of constant energy in the range 200hz to 3400hz. it is intended to detect both white and pink noise and can be used to detect line dis- connection during recording. it is recommend to use the constant energy mechanism in conjunction with the no-energy (vox) mechanism. the following tunable parameters control the operation of the constant-energy detector: const_nrg_det_time_count const_nrg_det_tolerance_time const_nrg_det_low_threshold const_nrg_det_high_threshold no energy (vox) the no-energy detector reports when the energy in the frequency range 200hz to 3400hz remains below a preprogrammed threshold for a prepro- grammed time-out. a programmable tolerance is allowed. it is recommend to use the no-energy (vox) mechanism in conjunction with the constant-en- ergy mechanism. the following tunable parameters control the operation of the no-energy (vox) mechanism: vox_det_energy_threshold vox_det_time_count vox_det_tolerance_time 2.3.4 full-duplex speakerphone the speakerphone feature lets the user commu- nicate through a telephone line, using the unit? speaker and the microphone instead of its hand- set. the speakerphone processes signals sent from the line to the speaker, and from the micro- phone to the line. it also performs the necessary switching, attenuation and echo cancellation on the signals present on the line/speaker. the ISD-T360SB speakerphone is simple to use; it requires no special hardware or training for the echo cancelers. the gain control is fully digital, which eliminates the need for analog gain con- trol hardware. the speakerphone features two types of echoes, the electrical echo (line or circuit) and the acoustic echo. the electrical echo is a result of an imperfect impedance match between the 4- to 2-wire interface (hybrid) and the line imped- ance. the electrical echo, relatively short term, has a transfer function that varies slowly. the second echo, the acoustic echo, is a line imped- ance returning from the speaker to the micro- phone. this echo is relatively long term, and its transfer function may vary quite quickly if anyone, or anything, moves in the room. both echoes must be canceled to achieve a high- quality hands-free system. for more details of the speakerphone tunable parameters refer to table 2-9 of the command description section. e 1 e 2 e 3 s 1 s 2 s 3 [e 1 ? e 3 ] < 100 ms [s 1 ? s 3 ] < 100 ms 100 < e i < 1680 ms 70 < s i < 1220 ms
2?oftware ISD-T360SB 2-15 isd speakerphone terminology send path the signal path from the microphone (near-end speaker) to the line (far-end listener). the micro- phone is the input port, and line-out is the output port of this signal path. receive path the signal path from the line (far-end speaker) to the loudspeaker (near-end listener). the line-in is the input port, and the speaker is the output port for this signal path. aec acoustic echo controller. the part in the speak- erphone algorithm that controls the echo in the sendpath. eec electric echo controller. the part in the speaker- phone algorithm that controls the echo in the re- ceive path. speakerphone modes of operation full-duplex (on) the speakerphone works in full-duplex mode, meaning both parties can speak and hear each other simultaneously. in this mode both the acoustic and electric echo controllers are ac- tive. the voicedsp processor tone detectors are not active in this mode. mute in this mode of operation, the speakerphone generates silence to the line. the near-end listen- er can hear the far-end speaker but not vice ver- sa. tone detectors are not active. hold during hold mode interrupts from both codecs are stopped. neither side can hear each other. restart in restart mode the speakerphone re-initializes it- self to the last speakerphone mode (full-duplex, transparent or mute). this mode should be used to resume the speakerphone operation after hold mode or when there is a significant change in the environmental conditions (e.g., parallel pickup) that may affect the speakerphone qual- ity. transparent while in transparent mode, the speakerphone works in full-duplex mode but without echo can- cellation. samples from the microphone are transferred to the line, and samples from the line are trans- ferred to the speaker, with no processing. this mode should be used only for tuning and testing the system. listen in listen mode the line is audible on the speaker, and the processor tone detectors are active. during listen mode, dialing with the gt com- mand and call progress is possible, since the busy and dial tone detectors are active. the following pseudo-code demonstrates how to make a call from speakerphone mode:
2?oftware ISD-T360SB 2-16 voice solutions in silicon figure 2-3: speakerphone pseudo code representation 2.3.5 speech synthesis speech synthesis is the technology used to cre- ate messages out of predefined words and phrases stored in a vocabulary. there are two kinds of predefined messages: fixed messages (voice menus in a voice-mail sys- tem) and programmable messages (time-and- day stamp, or the you have n messages an- nouncement in a dtad). a vocabulary includes a set of predefined words and phrases, needed to synthesize messages in any language. applications which support more than one language require a separate vocabu- lary for each language. international vocabulary support (ivs) ivs is a mechanism by which the voicedsp pro- cessor utilizes several vocabularies stored on an external storage device. ivs enables the isd- t360sb to synthesize messages with the same meaning, but in different languages, from sepa- rate vocabularies. ivs features multiple vocabularies stored on a single storage device. plug-and-play. the same microcontroller code is used for all languages. synthesized and recorded messages use the same voice compression algorithm to achieve equal quality. argumented sentences. (for example: you have messages .) while () { ev = wait_event() case ev of: skpr_button_pressed: if (speakerphone_on) { ssm 0 // put voicedsp in idle mode first_digit = true deactivate_digit_timeout_event() else ssm 1 // put voicedsp in full-duplex speakerphone mode digit_pressed: if (first_digit) { ssm 4 // enter listen mode first_digit = false } gt // dial the digit s // stop. note that after the s command // the voicedsp is still in speakerphone mode enable_digit_timeout_event() // to "guess" when dialing is completed. digit_timeout_event: ssm 1 // dialing is completed, go back to full-duplex mode deactivate_digit_timeout_event() }
2?oftware ISD-T360SB 2-17 isd auto-synthesized time-and-day stamp (driven by the voicedsp processor? clock). support for various language and sentence structures: one versus many. (for example: you have one message versus you have two messages .) none versus many. (for example: you have no messages versus you have two messages .) number synthesis (english ?ighty versus french quatre-vingt ). word order (english twenty-one versus german einundzwanzig ). days of the week (monday through sunday versus sunday through saturday). vocabulary design there are several issues, sometimes conflicting, which must be addressed when designing a vo- cabulary. vocabulary content if memory space is not an issue, the vocabulary could contain all the required sentences, each recorded separately. if memory space is a concern, the vocabulary must be compact; it should contain the mini- mum set of words and phrases required to syn- thesize all the sentences. the least memory is used when phrases and words that are common to more than one sentence are recorded only once, and the ivs tool is used to synthesize sen- tences out of them. a good combination of sentence quality and memory space is achieved if you take the ?om- pact?approach, and extend it to solve pronun- ciation problems. for example, the word twenty is pronounced differently when used in the sen- tences you have twenty messages and you have twenty-two messages . to solve this prob- lem, words that are pronounced differently should be recorded more than once, each in the correct pronunciation. vocabulary recording when recording vocabulary words, there is a compromise between space and quality. on one hand, the words should be recorded and saved in a compressed form, and you would like to use the best voice compression for that pur- pose. on the other hand, the higher the com- pression rate, the worse the voice quality. another issue to consider is the difference in voice quality between synthesized and record- ed messages (e.g., between time-and-day stamp and icms in a dtad environment). it is more pleasant to the human ear to hear both messages have the same sound quality. vocabulary access sometimes compactness and high quality are not enough. there should be a simple and flexi- ble interface to access the vocabulary ele- ments. not just the vocabulary but the code to access the vocabulary should be compact. when designing for a multi-lingual environment, there are even more issues to consider. each vo- cabulary should be able to handle language- specific structures and designed in a coopera- tive way with the other vocabularies so that the code to access each vocabulary is the same. when you use the command to synthesize the sentence monday. 12:30 p . m ., you should not care in what language it is going to be played back. ivs vocabulary components this section describes the basic concept of an ivs vocabulary, its components, and the rela- tionships between them.
2?oftware ISD-T360SB 2-18 voice solutions in silicon basic concepts an ivs vocabulary consists of words, sentences, and special codes that control the behavior of the algorithm which voicedsp processor uses to synthesize sentences. word table the words are the basic units in the vocabulary. create synthesized sentences by combining words in the vocabulary. each word in the vo- cabulary is given an index which identifies it in the word table. note that, depending on the language struc- tures and sentences synthesized, you may need to record some words more than once in the vo- cabulary. for example, if you synthesize the sen- tences: you have twenty messages and you have twenty-five messages , the word twenty is pronounced differently. they should, therefore, be defined as two different words. number tables the number tables allow you to treat numbers differently depending on the context. example 1: the number 1 can be announced as one as in message number one or as first as in first message . example 2: the number 0 can be announced as no as in you have no messages or as oh as in monday, eight oh five a . m . a separate number table is required for each particular type of use. the number table con- tains the indices of the words in the vocabulary that are used to synthesize the number. up to nine number tables can be included in a vocab- ulary. sentence table the sentence table describes the predefined sentences in the vocabulary. the purpose of this table is to make the microcontroller that drives the voicedsp processor independent of the lan- guage being synthesized. for example, if the flash and/or rom memory contain vocabularies in various languages, and the first sentence in each vocabulary means you have n messages, the microcontroller switches languages by issu- ing the following command to voicedsp proces- sor: sv , -select a new vocabulary the microcontroller software is thus independent of the grammar of the language in use. the sen- tences consist of words, which are represented by their indices in the vocabulary. sentence 0 all sentences but one are user defined. the voicedsp processor treats the first sentence in the sentence table (sentence 0) specially, to support time-and-day stamp. the processor as- sumes that the sentence is designed for both sys- tem time, and message time-and-day stamp announcement, and has one argument which is interpreted as follows: 0 system time is announced 1 the time-and-day stamp of the cur- rent message is announced. example 1: when the microcontroller sends the command: sas 0, 0 the system time and day is announced. example 2: when the microcontroller sends the command: sas 0, 1 the current message time-and-day stamp is announced. the following figure 2-4 shows the interrelation- ship between the three types of tables.
2?oftware ISD-T360SB 2-19 isd figure 2-4: the interrelationship between the word, the number, and the sentence tables control and option codes the list of word indices alone cannot provide the entire range of sentences that the voicedsp pro- cessor is able to synthesize. ivs control and op- tion codes send special instructions to control the speech synthesis algorithm? behavior in the processor. for example, if the sentence should announce the time of day, the voicedsp processor should be able to substitute the current day and time in the sentence. these control words do not repre- sent recorded words, rather they instruct the pro- cessor to take special actions. the ivs tool the ivs tool includes two utilities: 1. the dos-based ivs compiler 2. ivstool for windows. a windows 3.1/95 based utility the tools help create vocabularies for the voicedsp processor. they take you from design- ing the vocabulary structure, through defining the vocabulary sentences, to recording the vo- cabulary words. ivs compiler the ivs compiler runs on ms-dos (version 5.0 or later) and enables you to insert your own vocab- ulary, (i.e., basic words and data used to create numbers and sentences, as directories and files in ms-dos). the ivs compiler then outputs a bi- nary file containing that vocabulary. in turn, this information can be burned into an eprom or flash memory to be used by the voicedsp soft- ware. note the ivs data cannot be stored in eprom when semi-parallel flash is used (samsung or toshiba). ivs voice compression each ivs vocabulary can be compiled with ei- ther the 5.3 kbit/s, the 9.9 kbit/s or the 16.8kbit/s voice compression algorithm, or in pcm format. define the compression rate before compilation. the voicedsp processor automatically selects the required voice decompression algorithm when the sv command chooses the active vo- cabulary. sentence table number table word table you have opt_number control_sing messages five twenty you have messages message
2?oftware ISD-T360SB 2-20 voice solutions in silicon graphical user interface (gui) the ivs package includes a windows utility to as- sist the vocabulary designer to synthesize sen- tences. with this utility, you can both compose sentences and listen to them. how to use the ivs tool with the voicedsp processor the ivs tool creates ivs vocabularies, and stores them as a binary file. this file is burnt into a rom device or programmed into a flash memory de- vice using the inj (inject ivs) command. the voicedsp processor so (say one word) com- mand is used to select the required vocabulary. the sw (say words), so, ss (say sentence) and sas (say argumented sentence) commands are used to synthesize the required word or sen- tence. the typical vocabulary-creation process is as follows: 1. design the vocabulary. 2. create the vocabulary files (as described in detail below). use vistula for windows 3.1/95 to simplify this process. 3. record the words using any standard pc sound card and sound editing software, that can create .wav files. 4. run the ivs compiler to compress the .wav files, and compile them and the vo- cabulary tables into an ivs vocabulary file. 5. repeat steps 1 to 4 to create a separate ivs vocabulary for each language that you want to use. 6. burn the ivs vocabulary files into a rom (or flash memory) device. use the inj (in- ject ivs) command to program the data into a flash device. note the ivs data cannot be stored in eprom when semi-parallel flash is used (samsung or toshiba). once the vocabulary is in place, the speech syn- thesis commands of the voicedsp processor can be used to synthesize sentences. figure 2-5 shows the vocabulary-creation pro- cess for a single table on a rom or flash memory device. figure 2-5: creation of an ivs vocabulary .wav files number tables sentence table pc + sound card editor ivs compressed files (.vcd) ivs vocabulary files rom programmer rom .wav file editor ivstool for windows flash inj command .ini file compiler
2-21 2?oftware ISD-T360SB isd 2.4 voicedsp processor commands?uick reference table table 2-2: speech commands command description opcode hex source state result state command parameters return value name s/a description bytes description bytes ccio s con?ure codec i/o 34 reset, idle no change con?_value 1 none - cfg s con?ure voicedsp 01 reset no change con?_value 3 none - cmsg s create message 33 idle msg_open tag, num_of_ blocks 2+2 none - cmt s cut message tail 26 idle no change time_length 2 none - cvoc s check vocabulary 2b idle no change none - test result 1 dm s delete message 0a idle no change none - none - dms s delete messages 0b idle no change tag_ref, tag_mask 2 + 2 none - gcfg s get con?uration value 02 reset, idle no change none - version 1 gew s get error word 1b all states no change none - error word 2 gi s get information item 25 play, record, synthesis, t one_ generate, idle no change item 1 item value 2 gl s get length 19 idle no change none - message length 2 gms s get memory status 12 idle no change none - recording time left 2 gmt s get message tag 04 idle no change none - message tag 2 gnm s get number of messages 11 idle no change tag_ref, tag_mask 2 + 2 number of messages 2 gsw s get status word 14 all states no change none - status word 2 gt a generate tone 0d idle tone_ generate tone (single tone or dtmf) 1 none - gtd s get time and day 0e idle no change time_day_ option 1 time and day 2 gtm s get tagged message 09 idle no change tag_ref, tag_mask,dir 2+2+1 message found 1 gtune s get tunable parameter 06 idle, reset no change index 1 parameter_ value 2 init s initialize system 13 reset, idle idle none - none - inj s inject ivs data 29 idle no change n, byte 1 ...byte n 4 + n none - mr s memory reset 2a reset, idle no change none - none - p a playback 03 idle play none - none -
2-22 2?oftware ISD-T360SB voice solutions in silicon pa s pause 1c play, record, synthesis, tone_generate, idle* no change none - none - pdm s go to power- down mode 1a idle no change none - none - r a record message 0c idle record tag (message tag), compression_ rate 2 + 1 none - rdet s reset detectors 2c idle no change detectors_ reset_mask 1 none - res s resume 1d play, record, synthesis, tone_generate idle* no change none - none - rmsg s read message 32 idle, msg_open msg_open none - data 32 s s stop 00 all states but reset idle none - none - sas a say argumented sentence 1e idle synthesis sentence_n, arg 1 + 1 none - sb s skip backward 23 play, idle* no change time_length 2 none - sdet s set detectors mask 10 idle no change detectors_ mask 1 none - se s skip to end of message 24 play, idle* no change none - none - setd s set time and day 0f idle no change time_and_ day 2 none - sf s skip forward 22 play, idle* no change time_length 2 none - smsg s set message pointer 30 idle, msg_open msg_open num_of_ pages 2 none - smt s set message tag 05 idle no change message_tag 2 none - so a say one word 07 idle synthesis word_number 1 none - sps s set playback speed 16 play, synthesis, idle no change speed 1 none - ss a say sentence 1f idle synthesis sentence_n 1 none - table 2-2: speech commands (continued) command description opcode hex source state result state command parameters return value name s/a description bytes description bytes
2-23 2?oftware ISD-T360SB isd note: * command is valid in idle state, but has no effect. s = synchronous command a= asynchronous command note: * command is valid in idle state, but has no effect. s = synchronous command a = asynchronous command sv s set vocabulary type 20 idle no change type, id 1 + 1 none - sw a say words 21 idle synthesis n, word 1 ...word n 1 + n none - tune s tune parameters 15 idle, reset no change index, parameter_ value 1 + 2 none - vc s volume control 28 play, synthesis, idle, tone_generat e no change vol_level (increment/ decrement) 1 none - wmsg s write message 31 idle, msg_open msg_open data 32 none - table 2-3: speakerphone commands command description opcode hex source state result state command parameters return value name s/a description bytes description bytes gew s get error word 1b tone_generat e, idle no change none - error word 2 gsw s get status word 14 tone_generat e, idle no change none - status word 2 gt a generate tone 0d idle tone_ge nerate tone (single tone or dtmf) 1 none - rdet s reset detectors 2c idle no change detectors_res et_mask 1 none - s s stop 00 tone_generat e, idle idle none - none - sdet s set detectors mask 10 idle no change detectors_ mask 1 none - ssm s set speakerphon e mode 2f idle no change mode 1 none - vc s volume control 28 tone_generat e, idle no change vol_level (increment/ decrement) 1 none - table 2-2: speech commands (continued) command description opcode hex source state result state command parameters return value name s/a description bytes description bytes
2-24 2?oftware ISD-T360SB voice solutions in silicon 2.5 command description the commands are listed in alphabetical order. the execution time for all commands, when specified, includes the time required for the microcontrol- ler to retrieve the return value, where appropriate. the execution time does not include the protocol timing overhead, i.e., the execution times are mea- sured from the moment that the command is detected as valid until the command is fully executed. note: each command description includes an example application of the command. the examples show the opcode issued by the microcontroller and the response returned by the voicedsp processor. for com- mands which require a return value from the processor, the start of the return value is indicated by a thick vertical line. ccio con?ure codec i/o con?_value configures the voice sample paths in various states. it should be used to change the default voicedsp processor configuration. it is relevant only when two codecs are used and speakerphonr is connect- ed. the config_value parameter is encoded as follows: bit 0 loopback control. 0 loopback disabled (default). 1 loopback enabled. in the record state, the input samples are echoed back, un- changed (i.e., no volume control), to the same codec. bit 1 codec input control. 0 input is received via the line codec (default). 1 input is received via the speakerphone codec. bits 2? codec output control. 00 in play, idle, synthesis and tone_generate states, output is to both codecs. in record state, output is to the non-input codec (no volume control). if the loopback control bit is set, output in record state is to both codecs as well (default). 01 output in all states is to the line codec. 10 output in all states is to the speakerphone codec. 11 output in all states is to both codecs.
2-25 2?oftware ISD-T360SB isd bits 4? reserved. example cfg con?ure voicedsp con?_value configures the voicedsp processor in various hardware environments. it should be used to change the default processor configuration. the config_value parameter is encoded as follows: bits 0? memory type. 0000 no memory (default). 0001 a/dram. 0010 reserved. 0011 toshiba serial flash. 0100 samsung semi-parallel flash. 0101 toshiba semi-parallel flash. 0110 reserved. 0111 reserved. bits 4? number of installed memory devices. 00 1 (default) 01 2 10 3 11 4 bit 6-14 reserved. bit 15 echo cancellation control (for dtmf detection). 0 echo cancellation off (default). 1 echo cancellation is on during playback. ccio 01 byte sequence: microcontroller 34 01 voicedsp 34 01 description: enable loopback
2-26 2?oftware ISD-T360SB voice solutions in silicon echo cancellation improves the performance of dtmf detection during playback. echo cancellation can be turned on only with a system that can disable hw agc (if present) during playback. a system featuring hw agc, that cannot be controlled by the microcontroller (i.e., disabled or enabled), must not turn on this bit. bit 16 clock bit rate (in slave mode only). 0 one bit rate clock (default). 1 two bit rate clock. bit 17 codec configuration. 0 short-frame format (default). 1 long-frame format (guaranteed by design but not tested). bits 18?9 codec type. 00 16-bit linear (default). 01 ?law. 10 a-law. bit 20 codec interface mode. 0 master codec interface (default). 1 slave codec interface. bits 21?2 number and type of codecs 00 one single codec (default). 01 two single codecs. 10 one dual codec. 11 reserved. the codecs should be connected as follows: telephone line or equivalent - always connected as channel 0. speaker and microphone - connected as channel 1 in case of one dual codec (not applicable in slave mode), connected as channel 2 in case of two single codecs.
2-27 2?oftware ISD-T360SB isd bit 23 reserved. example cmsg create message tag num_of_blocks creates a new message with a message tag tag , allocates num_of_blocks 4-kbytes blocks for the new message, and sets the message pointer to the beginning of the message data. cmsg switches the voicedsp processor to the msg_open state. the memory space available for the message data is computed as follows: (127 x num_of_blocks ?2) x 32 bytes. once a message is open (the processor is in the msg_open state), the message pointer can be set to any position on a page (32 bytes) boundary within the message with the smsg command. modify the message contents with the wmsg command, and read with the rmsg command. the microcontroller must issue an s command to close the message and switch the voicedsp proces- sor to the idle state. if the memory is full, ev_memfull is set in the status word and no message is created. if the memory is not full but there is insufficient memory and the processor cannot allocate more memory space, ev_memlow is set in the status word and no message is created. example cfg 144013 byte sequence: microcontroller 01 14 40 13 voicedsp 01 14 40 13 description: con?ure the voicedsp to work with: single codec in slave mode and a-law compressed samples. data in short frame format and single bit rate interface. two serial toshiba flash devices. echo cancellation for dtmf detectiopn is on. cmsg 0101 0001 byte sequence: microcontroller 33 01 01 00 01 voicedsp 33 01 01 00 01 description: create a new message with a tag=0101, and allocate 1 block (4 kbytes) for its data.
2-28 2?oftware ISD-T360SB voice solutions in silicon cmt cut message tail time_length cut time_length units, in 10 ms segments, off the end of the current message. the maximum value of time_length is 6550. in case of silence, cut-time accuracy is 0.1 to 0.2 seconds (depends on compres- sion rate). note if time_length is longer than the total duration of the message, the ev_normal_end event is set in the status word and the message becomes empty but not deleted. use the dm (delete message), or dms (delete messages), command to delete the message. a compressed frame represents 21 ms of speech, thus the minimum meaningful parameter is 3, (i.e., a 30 ms cut). cmt 1 or cmt 2 have no effect.the cmt command can not be used on data messages example cvoc check vocabulary checks (checksum) if the ivs data was correctly programmed to the rom or flash device. if the vocabulary data is correct the return value is 1. otherwise the return value is 0. if the current vo- cabulary is undefined, err_invalid is reported. example dm delete message deletes the current message. deleting a message clears its message tag. deleting the current message does not cause a different message to become current. the current message is undefined. if, for example, you issue the gtm command to skip to the next message, the first message that is newer than the just deleted message is selected as the current message. the memory space released by the deleted message is immediately available for recording new mes- sages. cmt 02bc byte sequence: microcontroller 26 02 bc voicedsp 26 02 bc description: cut the last seven seconds of the current message. cvoc byte sequence: microcontroller 2b aa voicedsp 2b 01 description: check the current vocabulary. the voicedsp processor r esponds that the vocabulary is ok.
2-29 2?oftware ISD-T360SB isd example dms delete messages tag_ref tag_mask deletes all messages whose message tags match the tag_ref parameter. only bits set in tag_mask are compared i.e., a match is considered successful if: message tag and tag_mask = tag_ref and tag_mask where and is a bitwise and operation. after the command completes execution, the current message is undefined. use the gtm command to select a message to be the current message. the memory space released by the deleted message is immediately available for recording new mes- sages. example gcfg get con?uration value returns a sequence of one byte with the following information: bits 0? magic number, which specifies the voicedsp firmware version. dm byte sequence: microcontroller 0a voicedsp 0a description: delete current message. dms ffc2 003f byte sequence: microcontroller 0b ff c2 00 3f voicedsp 0b ff c2 00 3f description: delete all old incoming messages from mailbox number 2 in a system where the message tag is encoded as follows: bits 0?: mailbox id 8 mailboxes indexed: 0 to 7 bit 3: new/old message indicator 0?essage is old 1?essage is new bits 4?: message type 00?cm/memo 01?gm 10?all transfer message bits 6?5: not used note: the description of the tag is an example only. all bits of the tag are user-de?able.
2-30 2?oftware ISD-T360SB voice solutions in silicon example gew get error word returns the 2-byte error word. error word the 16-bit error word indicates errors that occurred during execution of the last command. if an error is detected, the command is not processed; the ev_error bit in the status word is set to 1, and the mwrqst signal is activated (set to 0). the gew command reads the error word. the error word is cleared during reset and after execution of the gew command. if errors err_command or err_param occur during the execution of a command that has a return value, the return value is undefined. the microcontroller must still read the return value, to ensure prop- er synchronization. the bits of the error word are used as follows: err_opcode illegal opcode. the voicedsp processor does not recognize the opcode. err_command illegal command sequence. the command is not legal in the current state. err_param illegal parameter. the value of the parameter is out of range, or is not appropriate for the command. err_comm microcontroller microwire communication error. gcfg byte sequence: microcontroller 02 aa voicedsp 02 01 description: get the voicedsp processor magic number. the voicedsp processor responds that it is version 1. 159876543 2 10 res res err_ invalid err_ timeout err_ comm res err_ param err_ command err_ opcode res
2-31 2?oftware ISD-T360SB isd err_timeout time-out error. depending on the voicedsp processor? state, more than 100 milliseconds elapsed be- tween the arrival of two consecutive bytes (for commands that have parameters). err_invalid command can not be performed in current context. example gi get information item returns the 16-bit value specified by item from one of the internal registers of the voicedsp processor. item may be one of the following: 0 the duration of the last detected dtmf tone, in 10 ms units. the return value is meaningful only if dtmf detection is enabled, and the status word shows that a dtmf tone was de- tected. 1 the duration of the last detected busy tone in 10 ms units. 2 the energy level of the samples in the last 10 ms. 3 the energy level of the samples, in the last 10 ms, that are in the frequency range de- scribed in figure 2-1. the return value is meaningful only if one of the tone detectors is enabled (bits 0,1 of the detectors mask; see the description of sdet command). the return value is unpredictable for any other value of item. example gew byte sequence: microcontroller 1b aa aa voicedsp 1b 00 02 description: get the voicedsp processor error word (typically sent after gsw when ev_error is reported in the status word). the voicedsp processor responds: err_opcode: gi 0 byte sequence: microcontroller 25 00 aa aa voicedsp 25 00 00 06 description: get the duration of the last detected dtmf tone. the voicedsp processor responds: 60 ms.
2-32 2?oftware ISD-T360SB voice solutions in silicon gl get length returns the length of the current message in multiples of 4 kbytes (blocks). the returned value includes the message directory information (64 bytes for the first block and 32bytes for every other block), the message data, and the entire last block of the message, even if the mes- sage occupies only a portion of the last block. since a memory block includes 4096 bytes, the returned length may be bigger than the actual message length by up to 4095 bytes. the minimum length of a message is one block. example gms get memory status returns the total remaining memory blocks as a 16bit unsigned integer. the estimated remaining re- cording time may be calculated as follows: time = (num_of_blocks x 4096 x 8) / ( compression_rate x 1000) this estimate assumes no silence compression: a real recording may be longer, according to the amount of silence detected and compressed. example gmt get message tag returns the 16-bit tag associated with the current message. if the current message is undefined, err_valid is reported. gl byte sequence: microcontroller 19 aa aa voicedsp 19 00 04 description: get the length of the current message. the voicedsp processor responds: 4 i.e., the message occupies 16384 (4 * 4096) bytes. gms byte sequence: microcontroller 12 aa aa voicedsp 12 00 28 description: return the remaining memory blocks. the voicedsp responds: 40 blocks.
2-33 2?oftware ISD-T360SB isd example gnm get number of messages tag_ref tag_mask returns the number of messages whose message tags match the tag_ref parameter. only bits set in tag_mask are compared, a match is considered successful if: message tag and tag_mask = tag_ref and tag_mask where and is a bitwise and operation. the tag_ref and tag_mask parameters are each two bytes; the return value is also two bytes long. see ?essage tag?on page 2-3 for a description of message-tag encoding. if tag_mask = 0, the total number of all existing messages is returned, regardless of the tag_ref value. example gsw get status word returns the 2-byte status word. status word the voicedsp processor has a 16-bit status word to indicate events that occur during normal opera- tion. the voicedsp processor asserts the mwrqst signal (clears to 0), to indicate a change in the status word. this signal remains active until the voicedsp processor receives a gsw command. the status word is cleared during reset, and upon a successful gsw command. gmt byte sequence: microcontroller 04 aa aa voicedsp 04 00 0e description: get the current message tag. in a system where the message tag is encoded as described in the dms command, the voicedsp processor return value indicates that the message is a new icm in mailbox number 6. gnm fffe 0003 byte sequence: microcontroller 11 ff fe 00 03 aa aa voicedsp 11 ff fe 00 03 00 05 description: get the number of messages which have bit 0 cleared, and bit 1 set, in their message tags. voicedsp processor responds that there are ?e messages which satisfy the request. 15 14 13 12 11 10 9 8 7 6 5 4 3 0 ev_ dtmf ev_ reset ev_ vox ev_ const_ nrg res ev_ memlow ev_ dialtone ev_ busy ev_ error ev_ memfull ev_ normal_ end ev_ dtmf_ end ev_ dtmf_ digit
2-34 2?oftware ISD-T360SB voice solutions in silicon the bits in the status word are used as follows: ev_dtmf_digit dtmf digit. a value indicating a detected dtmf digit. (see the description of dtmf code in the gt command.) ev_dtmf_end 1 = ended detection of a dtmf tone. the detected digit is held in ev_dtmf_digit. ev_normal_end 1 = normal completion of operation, e.g., end of message playback. ev_memfull 1 = memory is full. ev_error 1 = error detected in the last command. you must issue the gew command to return the error code and clear the error condition. ev_busy 1 = busy tone detected. use this indicator for call progress and line disconnection. ev_dialtone 1 = dial tone detected. use this indicator for call progress and line disconnection. ev_memlow 1 = not enough memory. (see cmsg command for further details.) ev_const_nrg 1 = a period of constant energy was detected. use this indicator for line disconnection. (see const_nrg_time_count in table 2-8.) ev_vox 1 = a period of silence (no energy) was detected on the telephone line. use this indicator for line dis- connection. (see vox_time_count in table 2-8.) ev_reset when the voicedsp processor completes its power-up sequence and enters the reset state, this bit is set to 1, and the mwrqst signal is activated (cleared to 0). normally, this bit changes to 0 after per- forming the init command. if this bit is set during normal operation of the voicedsp processor, it indi- cates an internal voicedsp processor error. the microcontroller can recover from such an error by re- initializing the system. ev_dtmf 1 = started detection of a dtmf tone.
2-35 2?oftware ISD-T360SB isd example gt generate tone tone generates the tone specified by the 1-byte tone parameter. the voicedsp state changes to tone_generate. the tone generation continues until an s command is received. a dtmf or a single frequency tone may be generated as shown: to generate a dtmf encode the bits as follows: bit 0 1 bits 1? dtmf code. where the dtmf code is encoded as follows: value (hex)dtmf digit 0 to 90 to 9 aa b* c# db ec fd bits 5? 0 to generate a single frequency tone encode the bits as follows: bit 0 0 bits 1-5 3?0 the value in bits 1? is multiplied by 100 to generate the required frequency (300hz?000hz). gsw byte sequence: microcontroller 14 aa aa voicedsp 14 00 40 description: get the voicedsp processor status word (typically sent after the mmrqst signal is asserted by the voicedsp processor which indicates a change in the status word). the voicedsp processor responds that the memory is full.
2-36 2?oftware ISD-T360SB voice solutions in silicon bits 6-7 0 the voicedsp processor does not check for the validity of the tone specification. invalid specification yields unpredictable results. example gtd get time and day time_day_option returns the time and day as a 2-byte value. time_day_option may be one of the following: 0 get the system time and day. 1 get the current message time-and-day stamp. any other time_day_option returns the time-and-day stamp of the current message. time and day are encoded as follows: bits 0? day of the week (1 through 7). bits 3? hour of the day (0 through 23). bits 8?3 minute of the hour (0 through 59). bits 14?5 00 the time was not set before the current message was recorded. 11 the time was set, i.e., the setd (set time of day) command was executed. note if the current message is undefined, and time_day_option is 1, an err_invalid error is reported. gt 20 byte sequence: microcontroller 0d 20 voicedsp 0d 20 description: generate a single-frequency 1600hz tone.
2-37 2?oftware ISD-T360SB isd example gtm get tagged message tag_ref tag_mask dir selects the current message, according to instructions in dir , to be the first, n th next or n th previous mes- sage which complies with the equation: message tag and tag_mask = tag_ref and tag_mask . where and is a bitwise and operation. dir is one of the following: 0 selects the first (oldest) message. -128 selects the last (newest) message. n selects the n th next message starting from the current message. ? selects the n th previous message starting from the current message. to select the n th message with a given tag to be the current message you must first select the first mes- sage that complies with the above equation, and then issue another gtm command with n ?1 as a parameter, to skip to the n th message. note to select the n th , or -n th , message with a given tag to be the current message you must first select the first message (dir=0), or the last message (dir=-128), that complies with the above equation, and then issue another gtm command with n-1 (for next message), or -n+1 (for previous message), as a parameter, to skip to the n th , or -n th , message respectively. \f a message is found, it becomes the current message and 1 (true) is returned. if no message is found, the current message remains unchanged and 0 (false) is returned. if dir is not 0, and the current message is undefined the return value is unpredictable. after the com- mand execution the current message may either remain undefined or changed to any existing message. the only exception is when the gtm command is executed just after the dm command. (see the dm command for further details.) to access the n th message, when n > 127, a sequence of gtm commands is required. gtd 1 byte sequence: microcontroller 0e 01 aa aa voicedsp 0e 01 e8 29 description: get the current message time-and-day stamp. the voicedsp processor responds that the message was created on the ?st day of the week at 5:40 a . m . the return value also indicates that the setd command was used to set the system time and day before the message was recorded. note: if the sas command is used to announce the time-and-day stamp, ?onday is announced as the ?st day of the week. for an external vocabulary, the announcement depends on the vocabulary de?ition (see the ivs users manual for more details).
2-38 2?oftware ISD-T360SB voice solutions in silicon example gtune get tune index gets the value of the tunable parameter identified by index (one byte) as the 2-byte value, parameter_value . this command may be used to read and identify the parameter value that was set to tune the voicedsp. if index does not point to a valid tunable parameter, err_param is set in the error word. the gtune command may be used in idle state or reset state. if tune command was not used to set the tunable parameters, then the gtune command will read the default parameter value. tables 2-4 to 2-11 describe the tunable parameters, their index numbers and their default values. example gtm ffce 003f 0 byte sequence: microcontroller 09 ff ce 00 3f 00 aa voicedsp 09 ff ce 00 3f 00 01 description: select the oldest of the new icms, in mailbox number 6, to be the current message, for a system where the message tag is encoded as described in the example for the dms command. the voicedsp processor returns a value indicates that there is such a message. the following pseudo-code demonstrates how to play all new icms in mailbox number 6. the messages are marked as old after being played: return_val = gtm(ffce, 003f, 00) /*get the oldest message with the de?ed tag*/ while (returnval == true) begin p() /* play */ message_tag = gmt() /* get message tag */ smt(fff7) /* mark the message as ?ld */ gtm(ffce, 003f, 01) /* get next message with the same tag */ end gtune 17 byte sequence: microcontroller 06 17 aa aa voicedsp 06 17 02 bc description: get the minimum period for busy detection comactspeech responds: 700 (7 seconds).
2-39 2?oftware ISD-T360SB isd init initialize system execute this command after the voicedsp processor has been configured (see cfg command). init performs a soft reset of the voicedsp as follows: initializes the message directory information. messages are not deleted. to delete the messages, use the dm and dms commands. sets the detectors mask to 0. sets the volume level that is controlled by the vc command, to 0. sets the playback speed to normal (0). switches to the idle state. initializes the tone detectors. the current message is undefined after init execution. the tunable parameters are not affected by this command. they are set to their default values only during reset. example inj inject ivs data n byte 1 . . . byte n injects vocabulary data of size n bytes to good flash blocks. this command programs flash devices, on a production line, with ivs vocabulary data. it is optimized for speed; all voicedsp processor detectors are suspended during execution of the command. use the cvoc command to check whether programming was successful. if there is not enough memory space for the vocabulary data, err_param is set in the error word, and execution stops. flash blocks that include ivs data can not be used for recording, even if only one byte of the block contains ivs data (e.g., if the vocabulary size is 4k + 100 bytes, two blocks of the flash are not available for message recording). example init byte sequence: microcontroller 13 voicedsp 13 description: initialize the voicedsp processor. inj 00000080 data byte sequence: microcontroller 29 00 00 00 80 vocabulary data voicedsp 29 00 00 00 80 echo of data description: inject 128 bytes of vocabulary data.
2-40 2?oftware ISD-T360SB voice solutions in silicon mr memory reset erases all memory blocks and initializes the voicedsp processor (does exactly what the init command does). bad blocks, and blocks which are used for ivs vocabularies, are not erased. this command can be issued in either reset or idle states. note when memory reset is used in reset state, it must be issued after the cfg command is issued, or the memory type and number of devices are not defined. in this case the result is unpredictable. note the command erases all messages and should be used with care. example p playback begins playback of the current message. the voicedsp processor state changes to play. when play- back is complete, the voicedsp processor sets the ev_normal_end bit in the status word, and acti- vates (clears to 0) the mwrqst signal. the state then changes to idle. playback can be paused with the pa command, and can be resumed later with the res command. playback can be stopped with the s command. if the current message is undefined, err_invalid is reported. example pa pause suspends the execution of the current gt, p, r, sas, so, sw, or ss command. the pa command does not change the state of the voicedsp processor; execution can be resumed with the res command. note dtmf and tone detectors remain active during pause. mr byte sequence: microcontroller 2a voicedsp 2a description: erase all memory blocks. p byte sequence: microcontroller 03 voicedsp 03 description: play the current message.
2-41 2?oftware ISD-T360SB isd example pdm go to power-down mode switches the voicedsp processor to power-down mode (see ?ower-down mode?on page 1-4 for details). sending any command while in power-down mode resets the processor detectors, and re- turns it to normal operation mode. note if an event report is pending (mwrqst is active) and not processed by the microcontroller prior to issuing the pdm command, the event is lost. example r record tag compression_rate records a new message with message tag tag and compression rate compression_rate . the voiced- sp processor state changes to record. the r command continues execution until stopped by the s command. recording can be paused with the pa command, and can be resumed later with the res command. if the memory becomes full, recording stops and ev_memfull is set in the status word. see ?essage tag?on page 2-3 for a description of message-tag encoding. the compression rate may be defined as 0 for pcm recording or either 1, 3, 6 for compression rates 5.3 kbits/sec, 9.9 kbits/sec, 16.8 kbits/sec respectively. see ?vcd?on page 2-10 for a description of the compression algorithm. note a time-and-day stamp is automatically attached to each message. before using the r command for the first time, use the setd command. failure to do so results in undefined values for the time-and- day stamp. example of a typical recording session: (icm) the microcontroller detects the first ring. (icm, ogm, memo) the microcontroller sends the r command. pa byte sequence: microcontroller 1c voicedsp 1c description: suspend playback of current message. pdm byte sequence: microcontroller 1a voicedsp 1a description: put the voicedsp processor in power-down mode.
2-42 2?oftware ISD-T360SB voice solutions in silicon example rdet reset detectors detectors_reset_mask resets the voicedsp processor tone and energy detectors according to the value of the detectors_reset_mask parameter. a bit set to 1 in the mask, resets the corresponding detector. a bit cleared to 0 is ignored. the 1-byte detectors_reset_mask is encoded as follows: bit 0 reset the busy and dial tone detectors. bits 1? reserved. must be cleared to 0. bit 4 reset the constant energy detector. bit 5 reset the no energy (vox) detector. bit 6 reset the dtmf detector. bit 7 reserved. must be cleared to 0. example r 000e 03 byte sequence: microcontroller 0c 00 0e 03 voicedsp 0c 00 0e 03 description: record a new icm in mailbox number 6 in a system where the message tag is encoded as described in the example of the dms command. the compression rate is de?ed as 9.9 kbits/sec. rdet 20 byte sequence: microcontroller 2c 20 voicedsp 2c 20 description: reset the vox detector.
2-43 2?oftware ISD-T360SB isd res resume resumes the activity that was suspended by the pa, sb, or sf commands. example rmsg read message returns 32 bytes of data from the current position of the message pointer, and advances the message pointer by 32 bytes. if the voicedsp processor was in the idle state, the command opens the current message, switches the voicedsp processor to the msg_open state, sets the message pointer to the beginning of the mes- sage data, and returns the 32 bytes of data. the microcontroller must issue an s command to close the message, and switch the voicedsp proces- sor to the idle state. if the current message is undefined, err_invalid is reported. trying to read beyond the end of the message sets the ev_normal_end event, and the voicedsp processor switches to the idle state. in this case, the return value is undefined and should be ignored. example s stop stops execution of the current command and switches the voicedsp processor to the idle state. s may be used to stop the execution of cmsg, smsg, wmsg, rmsg and all asynchronous commands. example res byte sequence: microcontroller 1d voicedsp 1d description: resume playback which was suspended by either the pa, sf or sb command. rmsg byte sequence: microcontroller 32 aa aa ... voicedsp 32 32 bytes of data description: read 32 bytes from the current message memory. s byte sequence: microcontroller 00 voicedsp 00 description: stop current activity (e.g., playback, recording) and change the voicedsp processor to idle state.
2-44 2?oftware ISD-T360SB voice solutions in silicon sas say argumented sentence sentence_n arg announces sentence number sentence_n of the currently selected vocabulary, and passes arg . sentence_n and arg are each 1-byte long. the voicedsp processor state changes to synthesis. when playing is complete, the voicedsp processor sets the ev_normal_end bit in the status word, and activates the mwrqst signal. the state then changes to idle. if the current vocabulary is undefined, err_invalid is reported. example sb skip backward time_length skips backward in the current message time_length units, in 0.2 second segments, and pauses mes- sage playback. the res command must be issue to continue playback. time_length is a 2-byte pa- rameter that can have any value up to 320 (64 seconds). the skip accuracy is five percent. sb is meaningful only in the play state. if the beginning of the message is detected during the execution of the sb command, execution ter- minates, the ev_normal_end bit in the status register sets, the mwrqst signal activates, and the pro- cessor switches to the idle state. if time_length is greater than 320, err_param is set in the error word. playback speed does not affect the behavior of this command. example sdet set detectors mask detectors_mask controls the reporting of detection of tones and energy detectors according to the value of the detectors_mask parameter. a bit set to 1 in the mask, enables the reporting of the corresponding de- tector. a bit cleared to 0 disables the reporting. disabling reporting of a detector does not stop or reset the detector. the 1-byte detectors_mask is encoded as follows: sas 00 03 byte sequence: microcontroller 1e 00 03 voicedsp 1e 00 03 description: announce the ?st sentence in the sentence table of the currently selected vocabulary with ? as the actual parameter. sb 0019 byte sequence: microcontroller 23 00 19 voicedsp 23 00 19 description: skip backwards ?e seconds from the current position in the message being played.
2-45 2?oftware ISD-T360SB isd bit 0 report detection of a busy tone. bit 1 report detection of a dial tone. bit 2-3 reserved. must be cleared to 0. bit 4 report detection of a constant energy. bit 5 report detection of no energy (vox) on the line. bit 6 report the ending of a detected dtmf. bit 7 report the start of a detected dtmf (up to 40 ms after detection start). example se skip to end of message this command is valid only in the play state. when invoked, playback is suspended (as for the pa com- mand), and a jump to the end of the message is performed. playback remains suspended after the jump. example sdet a3 byte sequence: microcontroller 10 b3 voicedsp 10 b3 description: set reporting of all voicedsp pr ocessor detectors, except for end-of-dtmf. se byte sequence: microcontroller 24 voicedsp 24 description: skip to end of current message.
2-46 2?oftware ISD-T360SB voice solutions in silicon setd set time and day time_and_day sets the system time and day as specified by the 2-bytes time_and_day parameter. the time_and_day parameter is encoded as follows: bits 0? day of the week (1 through 7). bits 3? hour of the day (0 through 23). bits 8?3 minute of the hour (0 through 59). bits 14?5 must be set to 1. if time_and_day value is not valid, err_param is set in the error word. example sf skip forward time_length skips forward in the current message time_length units, in 0.2 second segments, and causes message playback to pause. the res command must be issue to continue playback. time_length is a 2-byte parameter that can have any value up to 320 (64 seconds). the skip accuracy is five percent. this command is meaningful only in the play state. the res command must be issue to continue playback. if the end of the message is detected during execution of sf, execution of the command is terminated the ev_normal_end bit in the status word is set, the mwrqst signal is activated and the processor switches to the idle state. if time_length is greater than 320, err_param is set in the error word. playback speed does not affect the behavior of this command. example setd de09 byte sequence: microcontroller 0f de 09 voicedsp 0f de 09 description: set time and day to monday 1.30 a . m . (where monday is the ?st day of the week) sf 0019 byte sequence: microcontroller 22 00 19 voicedsp 22 00 19 description: skip forward ?e seconds from the current position in the message being played.
2-47 2?oftware ISD-T360SB isd smsg set message pointer num_of_pages sets the message pointer to num_of_pages x 32 bytes from the beginning of the current message da- ta. if the voicedsp processor was in the idle state, the command opens the current message and switch- es the voicedsp processor to the msg_open state. the microcontroller must issue an s command to close the message, and switch the voicedsp processor to the idle state. if num_of_pages x 32 is greater than the message length, ev_normal_end is set in the status word, the message pointer is set to the end of the message, and the voicedsp processor switches to the idle state. if the current message is undefined, err_invalid is reported. example smt set message tag message_tag sets the tag of the current message. the 2-byte message_tag can be used to implement mailbox functions by including the mailbox number in the tag, or to handle old and new messages differently by using one bit in the tag to mark the message as old or new. see ?essage tag?on page 2-3. to change the message tag, you should first get the tag using the gmt command, read the tag, mod- ify it, and write it back. note message tag bits can only be cleared. message tag bits are set only when a message is first created. if the current message is undefined, err_invalid is reported. example smsg 000a byte sequence: microcontroller 30 00 0a voicedsp 30 00 0a description: set the message pointer to 10 pages (320 bytes) from the beginning of the current message data. smt fff7 byte sequence: microcontroller 05 ff f7 voicedsp 05 ff f7 description: mark the current message as old in a system where the message tag is encoded as described in the example of the dms command. note that the voicedsp processor ignores bits in the tag which are set to 1; only bit 3 is modi?d in the message tag.
2-48 2?oftware ISD-T360SB voice solutions in silicon so say one word word_number plays the word number word_number in the current vocabulary. the 1-byte word_number may be any value from 0 through the index of the last word in the vocabulary. the voicedsp processor state changes to synthesis. when playback of the selected word has been completed, the voicedsp processor sets the ev_normal_end bit in the status word, and activates the mwrqst signal. the state then changes to idle. if word_number is not defined in the current vocabulary, or if it is an ivs control or option code, err_param is set in the error word. if the current vocabulary is undefined, err_invalid is reported. example sps set playback speed speed sets the speed of message playback as specified by speed. the new speed applies to all recorded messages and synthesized messages (only if synthesized using ivs), until changed by another sps com- mand. if this command is issued while the voicedsp processor is in the play state, the speed also changes for the message currently being played. speed may be one of 13 values, from ? to +6. a value of 0 represents normal speed. note a negative speed value represents an increase in speed, a positive value represents a decrease in speed. note the playback speed control is not applicable when the stored messages or the ivs data are not compressed (stored in pcm format). the change in speed is approximate, and depends on the recorded data. in any case, if i < j, play- back speed with parameter i is the same or faster than with parameter j. if speed is not in the ? to +6 range, err_param is set in the error word. example so 00 byte sequence: microcontroller 07 00 voicedsp 07 00 description: announce the ?st word in the word table of the currently selected vocabulary. sps fb byte sequence: microcontroller 16 fb voicedsp 16 fb description: set playback speed to ?.
2-49 2?oftware ISD-T360SB isd ss say sentence sentence_n say sentence number sentence_n of the currently selected vocabulary. sentence_n is 1-byte long. the voicedsp processor state changes to synthesis. if the sentence has an argument, 0 is passed as the value for this argument. when playing has been completed, the voicedsp processor sets the ev_normal_end bit in the status word, and activates the mwrqst signal. the state then changes to idle. if sentence_n is not defined in the current vocabulary, err_param is set in the error word. if the current vocabulary is undefined, err_invalid is reported. example ssm set speakerphone mode mode sets the speakerphone to the mode mode of operation. the command is valid when the voicedsp processor is in idle state. mode can be one of: 0 off deactivate the speakerphone, and return the voicedsp processor to normal operation mode. 1 on put the voicedsp processor in speakerphone mode and activate speak- erphone in full-duplex mode i.e., with full cancellation of both the acous- tic and the electrical echoes. tone detectors are not active. gains in the send and receive paths are set by the relevant tunable parameters. 2 transparent activate the speakerphone with no echo cancellation (this mode is used for system tuning). 3 mute activate the speakerphone, while generating silence to the line. the near-end-listener can hear the far-end-speaker, but not vice versa. tone detectors are not active. 4 listen the line is audible on the speaker. tone detectors are active. this mode is used for call generation. 5 reserved. 6 restart restart the current speakerphone mode. this mode differs from on; it does not require full initialization of the speakerphone. it should be used to resume the speakerphone operation after hold mode or to adjust to an environment change (e.g., parallel pickup). 7 hold stop the codec interrupts. neither side can hear each other. see ?ull-duplex speakerphone?on page 2-14 for more details. ss 00 byte sequence: microcontroller 1f 00 voicedsp 1f 00 description: announce the ?st sentence in the sentence table of the currently selected vocabulary.
2-50 2?oftware ISD-T360SB voice solutions in silicon note only commands that are specified in table 2-3, are active during all speakerphone modes (other than 0). example sv set vocabulary type type id selects the vocabulary table to be used for voice synthesis. the vocabulary type is set according to the 1-byte type parameter: 0 for compatibility only. 1 external vocabulary in rom. 2 external vocabulary in flash. 3-7 reserved. the host is responsible for selecting the current vocabulary, with sv command, before using an sas, so, ss or sw command. each external vocabulary table has a unique id which is part of the vocabu- lary internal header (see the ivs user? guide for more details). if type is 1 or 2, the voicedsp processor searches for the one byte id parameter in each vocabulary table header until a match is found. if the id parameter does not point to a valid ivs vocabulary, err_param is set in the error word. example sw say words n word 1 . . . word n plays n words, indexed by word 1 to word n . the voicedsp processor state changes to synthesis. on completion, the ev_normal_end bit in the status word is set, and the mwrqst signal goes low. the state then changes to idle. if one of the words is not defined in the current vocabulary, or if it is an ivs control or option code, or if n > 8, err_param is reported. ssm 01 byte sequence: microcontroller 2f 01 voicedsp 2f 01 description: put the voicedsp processor into speakerphone mode, and set the speakerphone to full-duplex mode. sv 02 03 byte sequence: microcontroller 20 02 03 voicedsp 20 02 03 description: select the vocabulary with vocabulary-id 3, which resides on a flash, as the current vocabulary.
2-51 2?oftware ISD-T360SB isd if the current vocabulary is undefined, err_invalid is reported. example tune tune index parameter_value sets the value of the tunable parameter identified by index (one byte) to the 2-byte value, parameter_value . this command may be used to tune the dsp algorithms to a specific data access arrangement (daa) interface, or to change other parameters. if you do not use tune, the voicedsp processor uses default values. if index does not point to a valid tunable parameter, err_param is set in the error word. note the tunable parameters are assigned with their default values on application of power. the init command does not affect these parameters. the following tables 2-4 to 2-11 describe the tunable parameters, their index numbers and their default values, grouped by their functionality. sw 02 00 00 byte sequence: microcontroller 21 02 00 00 voicedsp 21 02 00 00 description: announce the ?st word, in the word table of the currently selected vocabulary, twice. table 2-4: tunable parameters: voice compression and decompression (vcd) index parameter name description default 4 voice activity detection (vad): vad_sil_threshold prevents speech from being interpreted as silence. the silence detection algorithm has an adaptive threshold, which is changed according to the noise level. this parameter is, therefore, only the initial threshold level. legal values: 9216 to 13824 in 512 (6 db) steps. 11264 5 voice activity detection (vad): vad_sil_threshold_st ep de?es the adaptive threshold changes step. if this threshold is too low, the threshold converges too slowly. if it is too high, silence detection is too sensitive to any noise. legal values: 3 to 48. 12 6 voice activity detection (vad): vad_sil_burst_thresh old the minimum time period for speech detection, during silence. as this threshold increases, the time period interpreted as silence increases. if this threshold is too low, a burst of noise is detected as speech. if it is too high, words may be partially cut off. legal values: 1 to 3. 2
2-52 2?oftware ISD-T360SB voice solutions in silicon 7 voice activity detection (vad): vad_sil_hang_thresh old the minimum time period for silence detection, during speech. as this threshold increases, the time period interpreted as silence decreases. if this threshold is too low, words may be partially cut off. if it is too high, no silence is detected. legal values: 8 to 31. 15 8 voice activity detection (vad): vad_sil_enable silence compression control. 0 turns silence compression off. note: silence compression must be turned off when using aram for voice storage. otherwise the playback quality is unpredictable. 1 9 voice activity detection (vad): vad_energy_factor determines the energy level used to synthesize silence. for the default value, the energy levels of the synthesized silence and the recorded silence are the same. if you divide (multiply) the default value by two, the synthesized silence is 6 db less (more) than the level of the recorded silence. legal values: 1024 to 16384. 8192 70 sw automatic gain control (sw agc): swagc_enable sw agc control. 0 turns sw agc off. 1 11 sw automatic gain control (sw agc): swagc_factor determines the maximum gain that the sw agc algorithm may use. legal values: 0, 1, 2, 4, 8, 16, 32, 64, 128 note: value 0 means the the maximum gain is de?ed by the algorithm. 128 table 2-4: tunable parameters: voice compression and decompression (vcd) index parameter name description default
2-53 2?oftware ISD-T360SB isd table 2-5: tunable parameters: tone generation and message playback index parameter name description default 27 dtmf generation: dtmf_gen_twist_level a one-byte value that controls the twist level of a dtmf tone, generated by the gt command, by controlling the energy level of each of the two tones (low frequency and high frequency) composing the dtmf tone. the least signi?ant nibble (lsn) controls the low tone and the most signi?ant nibble (msn) controls the high tone. the energy level of each tone, as measured at the output of a tp3054 codec (before the daa) connected to the voicedsp processor is summarized in the following table: nibble value tone energy (db-volts) 0 0 1 ?7.8 2 ?4.3 3 ?2.9 4 ?2.4 5 ?2.0 6 ?1.9 7 ?1.85 8..15 ?1.85 the volume of the generated dtmf tone during measurements was 6. (tone_gen_level+vol_level = 6). for the default level, the high tone is ?4.3 dbv and the low tone is ?2.4 dbv, which gives a dtmf twist level of 1.9 db. the energy level of a single generated tone is the level of the low tone. 66 16 tone generation: tone_gen_level controls the energy level at which dtmf and other tones are generated. each unit represents 3 db. the default level is the reference level. for example, if you set this parameter to 4, the energy level is 6 db less than the default level. the actual output level is the sum of tone_gen_level and the vol_level variable, controlled by the vc command. the tones are distorted when the level is set too high. legal values: 0 tone_gen_level + vol_level 12. 6 21 vcd playback and voice synthesis: vcd_play_level controls the energy during playback and external voice synthesis. each unit represents 3 db. the default level is the reference level. for example, if you set this parameter to 4, the energy level is 6 db less than the default level. the actual output level is the sum of vcd_level and the vol_level variable, controlled by the vc command. speech is distorted when the level is set too high. legal values: 0 vcd_play_level + vol_level 12. 6
2-54 2?oftware ISD-T360SB voice solutions in silicon table 2-6: tunable parameters: dtmf detection index parameter name description default 17 energy level: dtmf_det_min_energ y minimum energy level at which dtmf tones are detected. if you divide (multiply) the value by 2, the detection sensitivity decreases (increases) by 3 db. legal values: 8 to 4096 32 24 echo canceler: dtmf_det_echo_dela y the near-echo delay in samples. the sampling rate is 8000hz (i.e., 125 ms per sample). legal values: 0 to 16. 4 26 twist level: dtmf_det_rev_twist controls the reverse twist level at which the voicedsp processor detects dtmf tones. while the normal twist is set at 8 db, the reverse twist can be either 4 db (default) or 8 db (if this parameter is set to 1). 0 60 sw agc: dtmf_det_agc_idle sw agc for dtmf in idle/record modes. when incrementing the tunable by 1, the dynamic range is increased by 3 db. legal values: 0 to 5. 0 61 sw agc: dtmf_det_agc_play software agc for play mode and tone generation modes. when incrementing the tunable by 1, the dynamic range increases by 3 db. legal values: 0 to 16. 3
2-55 2?oftware ISD-T360SB isd table 2-7: tunable parameters: tone detection index parameter name description default 18 dial tone: tone_det_time_count controls the duration of a tone before it is reported as a dial tone, in 10 msec units. the accuracy of the constant is ?0 ms. legal values: 0 to 65535. 700 19 busy and dial tone: tone_det_on_energy _ threshold minimum energy level at which busy and dial tones are detected as on (after 700hz ?tering). if you divide (multiply) the value by 2 you get about 3 db decrease (increase) in the threshold. the mapping between energy level and the parameter value is as follows (measured on the codec output when a 400hz tone was injected to the codec input): tunable value energy threshold (db-volts) 10 ?1.8 20 ?8.6 100 ?1.7 500 ?4.7 8000 ?.5 legal values: 0 to 65535. 160 20 busy and dial tone: tone_det_off_energy _ threshold maximum energy level at which busy and dial tones are detected as off (after 700hz ?tering). if you divide (multiply) the value by 2 you get about 3 db decrease (increase) in the threshold. the mapping between energy level and the parameter value is the same as for tone_on_energy_threshold legal values: 0 to 65535. 110 23 busy tone: busy_det_min_time minimum time period for busy detection, in 10 ms units. the accuracy of the constant is ?0 ms. legal values: 0 to 65535. 600 53 busy tone: busy_det_min_on_tim e minimum period considered as on period for busy tone detection. note that for weak signals: (?0 db and below) the maximum value is 12 (i.e., 120 ms minimum detection time). unit: 10 ms. accuracy is ?0 ms. legal values: 10 to 1000. 10 54 busy tone: busy_det_max_on_tim e maximum period considered as on for busy-tone detection. unit: 10 ms. accuracy is ?0 ms. legal values: 10 to 1000. 168
2-56 2?oftware ISD-T360SB voice solutions in silicon 55 busy tone: busy_det_min_off_tim e minimum period considered as off for busy-tone detection. unit: 10 ms. accuracy is ?0 ms. legal values: 5 to 1000. 7 56 busy tone: busy_det_max_off_ti me maximum period considered as on for busy-tone detection. unit: 10 ms. accuracy is ?0 ms. legal values: 5 to 1000. 122 57 busy tone: busy_det_verify_cou nt number of on/off cadences that must be detected prior to reporting busy-tone presence. legal values: 9 to 127. 9 58 busy tone: busy_det_tone_type speci?s the type of busy tone to detect: 1 ?wo cadences 2 ?hree cadences 3 ?oth two and three cadences 1 59 busy tone: busy_det_diff_thresh old the maximum allowed difference between two compared on or off periods. unit: 10 ms. legal values: 0 to 1000. 9 table 2-7: tunable parameters: tone detection index parameter name description default
2-57 2?oftware ISD-T360SB isd table 2-8: tunable parameters: energy detection index parameter name description default 10 silence (vox): vox_det_energy_thre shold this parameter determines the minimum energy level at which voice is detected. below this level, it is interpreted as silence. legal values: 1 to 32767. 12 12 silence (vox): vox_det_time_count this parameter, in units of 10 ms, determines the period of silence before the voicedsp processor reports silence. the accuracy of the constant is ?0 ms. legal values: 0 to 65535. 700 22 silence (vox): vox_det_tolerance_t ime controls the maximum energy-period, in 10 ms units, that does not reset the vox detector. legal values: 0 to 255. 3 47 constant energy: const_nrg_det_time_ count minimum elapsed time until the voicedsp processor reports constant energy level. units: 10 ms. accuracy: ?0 ms legal values: 1 to 65534 700 48 constant energy: const_nrg_det_ tolerance_time variations in constant energy, up to this time, do not reset the constant energy detector. units: 10 ms. legal values: 0 to 255 5 49 constant energy: const_nrg_det_low _ threshold determines the minimum energy level that is treated as constant energy. the minimum energy is calculated as follows: (1?/2 const_nrg_det_low_threshold ) * average_energy legal values: 1 to 16 1 50 constant energy: const_nrg_det_high _ threshold determines the maximum energy level that is treated as constant energy. the maximum energy is calculated as follows: (1 + 1/2 const_nrg_det_high_threshold ) * average_energy legal values: 0 to 16 1
2-58 2?oftware ISD-T360SB voice solutions in silicon table 2-9: tunable parameters: speakerphone index parameter name description default 31 acoustic echo canceler (aec): sp_aec_priority_bias controls the bias in priority between the send and receive paths. if send priority-bias is preferred, the value should be greater than zero. for no priority bias, the value should be zero. for priority bias for the receive path, the value should be negative. steps are 3 db each (e.g., +3 is 9 db bias for the send path, ? is 6 db bias for the receive path). legal values: ? to 4. 0 32 acoustic echo canceler (aec): sp_aec_coupling_ loss_threshold this parameter limits the acoustic return loss. its value (sp_aec_coupling_loss_threshold / 32767) is compared with the rms value of s out divided by the rms value of r in , during a single-talk event. the loop gain is decreased, if necessary, to control the tcl level. for sp_aec_coupling_loss_threshold = 32767 this loop is disabled. legal values: 0 to 32767. 2047 34 acoustic echo canceler (aec): sp_aec_lr_level controls the speakerphone gain from the microphone to the line-out. the total attenuation, or gain, depends on both of the analog gains and this value. the gain is: k * signal where: k = sp_aec_lr_level/4096. legal values: 0 to 16000. 14000 36 acoustic echo canceler (aec): sp_aec_clip_pos speci?s the positive peak-value at which the analog circuit of the line-out saturates. codec analog full scale corresponds to ?aw full scale values after expansion.assume that positive saturation occurs at amplitudes higher than those of a sine wave at x [dbm0]. the sp_aec_clip_pos value is set as: sp_aec_clip_pos = 32636 * 10 ((x ?3.17)/20) note: a sine wave with amplitude 4 * 8159 = 32636 corresponds to 3.17 dbm0. example: for x = ?.2761 dbm0, the value is: sp_aec_clip_pos = 32636 * 10 ((?.2761?3.17)/20) = 0.3371 * 32636 =11000; legal values: 0 to 32767. 16000 37 acoustic echo canceler (aec): sp_aec_clip_neg speci?s the negative peak value at which the analog circuit of the line-out saturates. codec analog full scale corresponds to ?aw full scale values after expansion.the value of sp_aec_clip_neg is set as shown for sp_aec_clip_pos, above. legal values: ?2768 to 0. ?6000
2-59 2?oftware ISD-T360SB isd 40 acoustic echo canceler (aec): sp_aec_enable enables/disables the acoustic echo controller. legal values: 0 (disable), 1 (enable). 1 43 acoustic echo canceler (aec): sp_aec_vox_hyst controls the hysteresis in near-talker detection. (the speakerphone state machine has a built-in hysteresis mechanism to prevent ?ctuations in the talker identi?ation process i.e., identifying the active side.) the value of this parameter is a dimensionless number, which should be evaluated during the tuning process for speci? hardware. larger values for the parameter correspond to a wider hysteresis loop. negative values increase the probability that the state machine remains in the last state. legal values: ?27 to 127. 10 45 acoustic echo canceler (aec): sp_aec_dtd_th controls the sensitivity of the system. low values correspond to high sensitivity, with a greater false alarm probability (i.e., an echo is considered a real talker). high values correspond to low sensitivity, with slower switching. this parameter is affected by the loop gain and the speci? hardware characteristics. legal values: 0 to 127. 73 35 electric echo canceler (eec): sp_eec_lr_level controls the speakerphone gain from the line-in to the speaker. the total attenuation, or gain, depends on both of the analog gains and this value. the gain is: k * signal where: k = (sp_eec_lr_level/4096) * (2 (6 + vol_level)/2 ) legal values: 0 to 400. 281 table 2-9: tunable parameters: speakerphone index parameter name description default
2-60 2?oftware ISD-T360SB voice solutions in silicon 38 electric echo canceler (eec): sp_eec_clip_pos speci?s the positive peak value at which the analog circuit of the speaker saturates. codec analog full scale corresponds to ?aw full scale values after expansion. the value of sp_eec_clip_pos is set as shown for sp_aec_clip_pos, above. legal values: 0 to 32767. 16000 39 electric echo canceler (eec): sp_eec_clip_neg speci?s the negative peak value at which the analog circuit of the line-out saturates. codec analog full scale corresponds to ?aw full scale values after expansion. the value of sp_eec_clip_neg is set as shown for sp_aec_clip_pos, above. legal values: ?2768 to 0. ?6000 41 electric echo canceler (eec): sp_eec_enable enables/disables the electrical echo controller. legal values: 0 (disable), 1 (enable). 1 44 electric echo canceler (eec): sp_eec_vox_hyst controls the hysteresis in far-talker detection. (the speakerphone state machine has a built-in hysteresis mechanism to prevent ?ctuations in the talker identi?ation process i.e., identifying the active side.) the value of this parameter is a dimensionless number, which should be evaluated during the tuning process for speci? hardware. larger values for the parameter correspond to a wider hysteresis loop. negative values increase the probability that the state machine remains in the last state. legal values: ?27 to 127. 10 46 electric echo canceler (eec): sp_eec_dtd_th controls the sensitivity of the system. low values correspond to high sensitivity, with a greater false alarm probability (i.e., an echo is considered a real talker). high values correspond to low sensitivity, with slower switching. this parameter is affected by the loop gain and the speci? hardware characteristics. legal values: 0 to 127. 82 33 attenuation: sp_block_level controls the maximum attenuation level of the speakerphone suppressors. it affects the speakerphone stability and its subjective quality. the maximum attenuation is calculated according to: sp_block_level/2 28 legal values: 550 to 32000. 10922 42 tone generation: sp_tone_gen_ level controls the energy level at which dtmf, and other tones, are generated to the line (codec 0) while the speakerphone is active. each unit represents 3 db. legal values: 0 sp_tone_gen_level 10. note: the energy level at which the tones are generated to the speaker (codec 1) while the speakerphone is active, is controlled by the tone_gen_level tunable parameter and the vol_level. 6 table 2-9: tunable parameters: speakerphone index parameter name description default
2-61 2?oftware ISD-T360SB isd table 2-10: tunable parameters: memory support index parameter name description default 62 memory device size: num_of_blocks_in_m em de?es the nubber of blocks (each block is of 4096 bytes) in every memory device (flash or aram/dram). the number and type of connected devices are de?ed by the cfg command. flash device size (mbits) number of blocks value 4 128 8 256 16 512 aram/dram device size (mbits) number of blocks value 16 508 128 63 memory size for testing num_of_blocks_for_ test de?es the nubber of blocks (each block is of 4096 bytes) in every memory device (flash or aram/dram) for production line testing purposes. the number should be small to minimize testing time during the production sequence. h owever, the number of blocks should be larger than the number of expected bad blocks in the memory device. in case of value=0, no productiontest is performed. in any case other than value= 0, the number of blocks is de?ed by the parameter value, and a production testing cycle is performed after reset. legal values: 0 to 128. note: if power fails during production testing cycle, the memory status is unpredicted. the memory device should be replaced and the production test should be repeated. 0 64 aram quality level: max_defect_nibbles_i n_block de?es the maximum allowed bad nibbles in aram block (each block is of 8192 nibbles). a nibble (4 bits) is considered bad if any bit is defected. if the number of bad nibbles in a block exceeds the maximum allowed value, the block is marked as bad block and is not used for voice storage. legal values: 0 to 255. 0
2-62 2?oftware ISD-T360SB voice solutions in silicon example vc volume control vol_level controls the energy level of all the output generators (playback, tone generation, and voice synthe- sis), with one command. the resolution is ? db. the actual output level is composed of the tunable level variable, plus the vol_level . the valid range for the actual output level of each output generator is defined in table 2-5. for example, if the tunable variable vcd_level (parameter number 21) is 6, and vol_level is ?, then the output level equals vcd_level + vol_level = 4. example table 2-11: tunable parameters: codec support (samples) index parameter name description default 65 channel 0 delay: cfrd0 the delay of codec channel 0 from frame synch 0 (cfs0) to start of valid data. legal values: 0 to 255 1 66 channel 1 delay: cfrd1 the delay of codec channel 1 from frame synch 0 (cfs0) to start of valid data. legal values: 0 to 255 10 67 channel 2 delay: cfrd2 the delay of codec channel 2 from frame synch 0 (cfs0) to start of valid data. legal values: 0 to 255 10 68 frame synch delay: cfsd the delay of frame synch 1 (cfs1) from frame synch 0 (cfs0). legal values: 0 to 255 10 69 data valiid delay: cfet the delay between frame synch 0 (cfs0) to end of valid data of all channels. legal values: 0 to 255 18 tune 17 02bc byte sequence: microcontroller 15 17 02 bc voicedsp 15 17 02 bc description: set the minimum period for busy detection to 700 (7 seconds). vc 04 byte sequence: microcontroller 28 04 voicedsp 28 04 description: set the volume level to vcd_level + 4.
2-63 2?oftware ISD-T360SB isd wmsg write message data writes 32 bytes of data to the current position of the message pointer, and advances the message pointer by 32 bytes. if the voicedsp processor is in the idle state, the command opens the current message, switches the voicedsp processor to the msg_open state, sets the message pointer to the beginning of the mes- sage data, and writes the 32 bytes of data. to add data at the end of an existing message, issue the smsg command to the last page of the mes- sage. issue the wmsg command with a buffer consisting of 32 ff bytes (this has no effect on the cur- rent data in the page). a subsequent wmsg command adds a new block to the message, and writing continues at the beginning of the new block. the microcontroller must issue an s command to close the message and switch the voicedsp proces- sor to the idle state. note when updating an existing message, bits can only be cleared, but not set. if the current message is undefined, err_invalid is reported. example wmsg 32 bytes byte sequence: microcontroller 31 32 bytes of data to write voicedsp 31 echo 32 bytes of data description: write 32 bytes in the message memory.
2?oftware ISD-T360SB 2-64 voice solutions in silicon
3?chematic diagrams ISD-T360SB 3-1 isd chapter 3?schematic diagrams 3.1 application information the following schematic diagrams for a voicedsp processor reference design unit. this reference design includes three basic clusters: an 80c51 microcontroller. voicedsp processor cluster, including a tp3054 codec, and an isdt360sb controlling a flash device. user interface that includes one 16-digit lcd, and a 16-key (4 x 4) keypad.
3?chematic diagrams ISD-T360SB 3-2 voice solutions in silicon
4?hysical dimensions ISD-T360SB 4-1 isd chapter 4?physical dimensions figure 4-1: 80-pin plastic quad flat package, top and bottom?ype: metric pqfp, 14x14 body 1. all dimensions are in millimeters. all dimensions and tolerances conform to ansi y14.5-1982.
4?hysical dimensions ISD-T360SB 4-2 voice solutions in silicon figure 4-2: 80-pin plastic quad flat package, side?ype: metric pqfp, 14x14 body table 4-1: packaging dimensions symbol min. nom. max. a 2.82 3.00 a 1 0.10 0.15 0.25 a 2 2.55 2.67 2.75 d 17.20 bsc. d 1 14.00 bsc. d 2 12.35bsc. z d 0.825 ref. e 17.20 bsc. e 1 14.00 bsc. e 2 12.35 bsc. z e 0.825 ref. l 0.73 0.88 1.03 n 80 e 0.65 bsc. b 0.22 0.38 b 1 0.22 0.30 0.33 ccc 0.12
part no. 2201298d5008 2045 hamilton ave. san jose, california 95125-5904 tel: 408/369-2400 fax: 408/369-2422 http://www.isd.com important notices the warranty for each product of isd (information storage devices, inc.), is contained in a written warranty which governs sale and use of such product. such warranty is contained in the printed terms and conditions under which such product is sold, or in a separate written warranty supplied with the product. please refer to such written warranty with respect to its applicability to certain applications of such product. these product may be subject to restrictions on use. please contact isd, for a list of the current additional restrictions on these product. by purchasing these product, the purchaser of these product agrees to comply with such use restrictions. please contact isd for clarification of any restrictions described herein. isd, reserves the right, without further notice, to change the isd chipcorder product specifications and/or information in this document and to improve reliability, functions and design. isd assumes no responsibility or liability for any use of the isd chipcorder product. isd conveys no license or title, either expressed or implied, under any patent, copyright, or mask work right to the isd chipcorder product, and isd makes no warranties or representations that the isd chipcorder product are free from patent, copyright, or mask work right infringement, unless otherwise specified. application examples and alternative uses of any integrated circuit contained in this publication are for illustration purposes only and isd makes no representation or warranty that such applications shall be suitable for the use specified. the 100-year retention and 100k record cycle projections are based upon accelerated reliability tests, as published in the isd reliability report, and are neither warranted nor guaranteed by isd. information contained in this isd chipcorder data sheet supersedes all data for the isd chipcorder product published by isd prior to december, 1998. this data sheet and any future addendum to this data sheet is (are) the complete and controlling isd chipcorder product specifications. in the event any inconsistencies exist between the information in this and other product documentation, or in the event that other product documentation contains information in addition to the information in this, the information contained herein supersedes and governs such other information in its entirety. copyright?1998, isd (information storage devices, inc.) all rights reserved. isd is a registered trademark of isd. chipcorder is a trademark of isd. all other trademarks are properties of their respective owners.

▲Up To Search▲

Price & Availability of ISD-T360SB

	To Download ISD-T360SB Datasheet File
If you can't view the Datasheet, Please click here to try to view without PDF Reader .