Spanish Text To Speech User`s Manual
Contents
1. Beginning of the file G B Great Britain U S A United States of America End of the file Note After modifications the abbreviation lexicon must be reloaded in memory 15 Depending on the platform OKI SCP middle ware Spanish Text To Speech Users Manual 6 APPENDIX A List of ASCII codes translated 6 1 7 bits ASCII characters PoC em imma O ee o Am separatoroword I aaa ie ignored O se Sepai otword Emma exclamation point mark pause xa ve FF E O E O AAA zz E lt lt A im punctuation pause or separator of phone number punctuation pause or DEE of phone number sign estrella estrella Y punctuation pause or decimal comma coma punctuation pause or hyphen or sign menos OKI SCP middle ware Spanish Text To Speech Users Manual ETS TY ME date separation 48 Po digit zero time separation OKI SCP middle ware Spanish Text To Speech Users Manual xsr tente RO ete S S OKI SCP middle ware Spanish Text To Speech Users Manual ater SOS esate a mater C aw o mater SOS a o y matter SOS T matter es tio ease A e E MO A E a AA os OKI SCP middle ware Spanish Text To Speech Users Manual 6 2 8 bits ASCII characters Recognised as translated by Decimal ASCII IBM extended code Character Character translated by j te is 155 ignored gt ignored ignored small letter2. Buenos dias T 30 se or T 30 Dupont 3 2 3 3 Voice volume Specifies the loudness of voice table 3 1 Code format n From 100 min to 0 max The default value is 0 PD Return to default setting 3 2 3 4 Pause Control This control allows a pause in the text table 3 1 _ Code format p1000ms 1000 millisecond pause in the text p1s 1 second pause in the text p2mn 2 minute pause in the text Example Buenos dias p1000ms m Buenos dias p1s mi Buenos dias p2mn mi 3 2 3 5 Modulated sound output Output modulated sounds OKI SCP middle ware Spanish Text To Speech Users Manual table 3 1 T Pa ea T B4 Chime 2 rising tone short short short long 3 B3 Chime 1 short long I B5 Chime 3 falling tone short short short long 3 2 4 Command Specification Commands are interrupting processes that are completely asynchronous with MSM7630 s internal processes Synthesis Stop pause and restart are provided by commands Commands are invalid in text to speech synthesis used primarily to control the sequence of speech synthesis Commands are allocated to control codes below 0x20 3 2 4 1 Stop Stops the current text to speech synthesis process table 3 1 Code format AC 03H Stop the current Text to Speech synthesis process The stop command causes MSM7630 to discard all text captured so far during synthesis including
3. 12 13 14 39 will be pronounced un dos mas en tres mas un cuatro es igual tres nueve 10 9 See chapter Control code specification 10See chapter Control code specification for a scientific pronunciation OKI SCP middle ware Spanish Text To Speech Users Manual 13 12 1 will be pronounced un tres un dos es igual un Examples 02 123454 will be pronounced cero dos pause un dos tres cuatro cinco cuatro 12 2345 456 will be pronounced un dos pause dos tres cuatro cinco cuatro cinco seis ab12x will be pronounced aa bee doce equis When there is a combination of digits and letters the system pronounce the letters letter by letter and the numbers by group of more 3 digits 12 13 14 39 will be pronounced un dos mas en tres mas un cuatro es igual tres nueve 2 13 12 1 will be pronounced un tres un dos es igual un 13 11 See chapter Control code specification for a scientific pronunciation 12 See chapter Control code specification for a scientific pronunciation 13 See chapter Control code specification for a scientific pronunciation OKI SCP middle ware Spanish Text To Speech Users Manual 5 User lexicons The characters in the user lexicon files must be coded in IBM extended ASCII 5 1 Exceptions lexicon 5 1 1 Using the lexicon The exceptions lexicon permits to change the pronunciation of a word or a group of consecutive words Some spanish and foreign words which are not pronounced in accordance with the b
4. small letter ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored OKI SCP middle ware Spanish Text To Speech Users Manual 2000 ignored I maa E aaaea T H o wa mona pora H o ea E O owa sn oaao T aaa E sian mins O mae ere ignored we 2 _aveston marc sign libra ignored ignored ignored a small letter i small letter 6 small letter small letter small letter capital letter ordinal symbol ordinal symbol question mark ignored ignored ignored x ignored exclamation mark ignored M ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored ignored Va OKI SCP middle ware Spanish Text To Speech Users Manual ignored z small letter ignored ignored ignored ignored ignored ignored ignored ignored N ignored ignored X ignored small letter ignored ignored ignored 00 ignored LO ignored ignored ignored ignored ignored ignored ignored i ignored ignored ignored ignored ignored ignored N ignored ignored aN ignored z o Z r z lt lt x lt al ignored ignored ignored OKI SCP middle ware Spanish Text To S
5. 04H to each sentence of text and sends the sentence to MSM7630 The host then must not send further text or Level 1 control codes until MSM7630 returns the synthesis termination code MSM7630 will return the synthesis termination code when output of synthesized sound ends After the synthesis termination code has been returned the host can immediately send the next text Fig 3 1 shows the sequence when return of synthesis termination codes has been specified and fig 3 2 shows the format of the synthesis termination code OKI SCP middle ware Spanish Text To Speech Users Manual g 0 Error location is 2 Byte binary data Error Location FFFFH normal termination not equal FFFFH indicates location where text analysis failed as number of bytes from start of text or from Error Code previous D Error code is data that indocates the coause of the error It will be FFFFH for normal termination fig 3 2 3 1 2 Exception Dictionary Read Mode In this mode an exception dictionary created by a utility that runs on the host is downloaded into the devices An exception dictionary is not appended to the previously sent user dictionary but entirely overwrites it An exception dictionary that has been sent cannot then be read 3 1 2 1 Dictionary transfer procedure for serial and microcontroller interfaces After the host has specified exception dictionary read mode refer to Control Codes Commands 1 Level 1 it will receive an ACK 06H
6. 1 bits eie e a Tc le 31 62 8 bits ASCII O Sas 35 A OKI SCP middle ware Spanish Text To Speech Users Manual 1 Introduction The Spanish Text To Speech system correctly synthesises the majority of Spnish texts It is sometimes necessary however to modify the text to make it compatible with the constraints given in the following paragraphs before submitting it to the Text To Speech process al OKI SCP middle ware Spanish Text To Speech Users Manual 2 User interface description Data transmission receipt between MSM7630 and the host processor is called the user interface Section of interface type is determined by the settings of the configuration register explained below Data means text data dictionary data and control codes 2 1 Reading the configuration register When MSM7630 starts up it reads external configuration register values and makes user interface and other environment settings The user interface to be used is determined by the configuration register value see Therefore the serial port and parallel port cannot be used in parallel table 2 1 Register Value interface 000 2400bps serial port 001 4800bps serial port 010 9600bps serial port 011 19200bps serial port 100 Micro controller interface The configuration register is connected to pins D 26 24 Pull up 10K register gives register value 1 also Pull down 10K register gives value 0 when the bus capacitance is 100pF Determine the value
7. code from MSM7630 and then will send the exception dictionary After MSM7630 receives the exception dictionary it performs a BCC check and based on the result sends a termination response of ACK 06H for normal termination or NACK 15H for abnormal termination After it sends the termination response MSM7630 will automatically transfer to its default operating mode text to speech synthesis mode 3 1 2 2 Time out In exception dictionary read mode MSM7630 will monitor the time interval between character transmissions When the interval timer times out about one second MSM7630 will transfer to text to speech synthesis mode It will not inform the host OKI SCP middle ware Spanish Text To Speech Users Manual HOST SCP Specify user dictionary read mode User dictionary read mode Dictionary Data ACK 06H NACK 15H Text to Speech synthesis mode fig 3 1 Note The BCC code 1 byte is for the exclusive OR of all data in the dictionary managemant table and the dictionary Data Length Dictionary Management Table and Dictionary BCC code note fig 3 2 3 1 3 Hardware sound output busy signal Busy signal should be given while sound output Busy signal is active low level OKI SCP middle ware Spanish Text To Speech Users Manual MSM7630 UPORT 50mS fig 3 1 OKI SCP middle ware Spanish Text To Speech Users Manual 3 2 Control Codes Specifications Control codes are sent by the host to con
8. expressions No information value c To pronounce Commercial No information value expressions t To pronounce telephone numbers No information value To pronounce roman numbers to enable and to disable 3 2 2 1 Usual pronunciation u This control restores the default mode The Control Name value is u there is no Control Information value Example 3 2 2 2 Scientific pronunciation s This control permits to pronounce the scientific expressions with the characters 1 the numeric value into a word or an expression like a number example 124 13 14 39 will be pronounced doce mas treice mas catorce es igual treinta y nueve always the minus sign example 13 12 1 and 13 12 1 will be pronounced treice menos doce es igual un The Control Name is s there is not Control Information To disable this control it is necessary to use an other control as usual commercial date or telephone because these control codes are exclusive Example 3 2 2 3 Commercial pronunciation c This control is not supported by this version 3 2 2 4 Pronunciation of dates d This control is not supported by this version 3 2 2 5 Pronunciation of telephone numbers t This control permits to pronounce the spanish telephone numbers like 10 12 699 551 Example OKI SCP middle ware Spanish Text To Speech Users Manual 10 12 will be pronounced diez pause doce and not el diez de doce The Control
9. interval from the rise of RTS to the fall of the start bit 2 2 2 Micro controller Interface When a micro controller interface is selected by the configuration register when register value is set to 100 the data transmit receive specification is as follows 8 Bit data port PD Status PIBF POBF Control PCS PA PWR PRD table 2 1 PCS PA PWR PRD Operation 1xxx Not operating 0010 PIBF POBF output PD high impedance For example to access from a host CPU connect as shown in the falling diagram fig 2 1 In the above case PIBF write buffer bit and POBF read buffer bit are connected wire OR to data port bits 7 and O respectively so the relation between address status and data is as follows o OKI SCP middle ware Spanish Text To Speech Users Manual Address Data 8bit xxx0 PIBF don t care POBF xxx1 parallel data fig 2 2 The data transfer process is as follows The xxx indicates a MSM7630 parallel port address Read xxx0 Address Read xxx0 Address Status Status Status Check no Port Busy Port Busy Bito 1 PIBF 1 yes E POBF 0 yes pce Write xxx1 Address Data Write xxx1 Address Data when receiving when synthesis termination code reply specified fig 2 3 For a parallel port when a synthesis termination code replay is specified the termination code might be missed unless the port is polled until a sentence has been transferred and the termination code
10. of each register so that the bus will stabilize within 18micro second 2 2 Individual Interface description o OKI SCP middle ware Spanish Text To Speech Users Manual 2 2 1 Serial port interface When a serial port interface is selected by the configuration register when register value is set to 000 001 010 or 011 the data transmit receive specification is as follows Data Format 8bit no parity 1stop bit Transfer Rate Selectable from 2400 4800 9600 or 19200bps Busy Control RTS Control The diagram below shows a serial port interface example fig 2 1 Be sure that the ports have sufficient drive capability The transmit receive process from the host is as follows Check Status no OK yes Transfer Data Transfer Data when receiving when synthesis termination code reply specified fig 2 2 The RTS pin will output 0 during reset and immediately after its release When the serial port cannot accept data or in other words when the serial port buffer 1Kbyte has become full the RTS pin output will change to 1 When the serial port can accept data the RTS pin will output 0 NI OKI SCP middle ware Spanish Text To Speech Users Manual Because RTS is controlled by software tens of clock may pass from output of the stop bit until RTS rises However RTS is set to become invalid when 128 bytes remain in the receive buffer so there will be no worry about overrun There is no standard time
11. 4 5 Punctuation Punctuation plays an important part in the texts analysed by the system It is necessary to put a space character just after the punctuation mark 4 5 1 Listof pronunciations recognised by the system and their effects table 4 1 pe rising small 4 5 2 Automatic breaks If a sentence contains too many words or too many characters without punctuation the system automatically inserts a full stop 4 5 3 Full stop A is always considered as a punctuation mark if it is not proceeded by an abbreviation or by anumber 4 6 Acronyms and abbreviations 4 6 1 List of acronyms and abbreviations of the system 6 See chapter Sentence 7 See chapter Numeration OKI SCP middle ware Spanish Text To Speech Users Manual The system does not deal with acronyms It will be try to pronounce the acronym as normal word 4 6 2 List of acronyms and abbreviations of the user List of abbreviations See the ABREVIAC RGS ASCII file Adding or modifying an abbreviation See the chapter Abbreviations lexicon Note At the end of the sentence if the last point is a full stop it must be separated from the abbreviation by a space character 4 7 Numeration 4 7 1 Numbers e Integers Examples 12 will be pronounced menos doce 123343 or 123 343 will be pronounced ciento veintitres mil trecientos cuarenta y tres 1912 will be pronounced mil novecientos doce 123 343 567 will be pronounced ciento veintitre
12. Name is t there is not Control Information To disable this control it is necessary to use an other control usual codes are exclusive Example scientific commercial date because these control 3 2 2 6 Pronunciation of roman numbers R This control permits to pronounce the roman numbers The roman numbers are composed with the capital letter I V X L C D M example IV will be pronounced cuatro The Control Name is R the Control Information is to enable and to disable The control usual permits also to disable it Example 3 2 3 Level 3 Control Code Level 3 control codes can be inserted anywhere between words in the text not just between sentence They primarily voice quality enabling fine control of voice quality for each word table 3 1 Pitch modification 2 Speed rate modification a Pause control a S Modulated sound output 3 2 3 1 Pitch modification This control permits to change the pitch in the text table 3 1 Code format n From 100 low to 100 high The default value is 0 HD Return to default setting Example Buenos dias H 10 se or Lopez H 10 N OKI SCP middle ware Spanish Text To Speech Users Manual 3 2 3 2 Speed rate modification This control permits to modify the speech rate of the text table 3 1 Code format n From 100 slow to 20 fast The default value is 0 TD Return to default setting Example
13. OKI SCP middle ware Spanish Text To Speech Users Manual OKI OKI middle ware for Speech Control Processor Spanish Text To Speech User s Manual 30 March 2000 Version 1 2 OKI SCP middle ware Spanish Text To Speech Users Manual 1 30 Mar 2000 modify of speed rate range NI OKI SCP middle ware Spanish Text To Speech Users Manual A A O O 5 2 User interface description isrcociaiia ie a ci 6 2 1 Reading the Configuration register esec eee eee 6 22 Individual Interface descripta 6 2 2 1 Serial port Tain eT 7 2 2 2 Micro controller Interact 8 2 2 3 MSM7630 Start up SequenCe oooococcccccccconcconccnnnnnnononononannnnnncononennnnnncnnnnnno nena nn cnnnnanonenannanancncnnns 9 3 Text To Speech program specification ommmccnnnmmen renace 12 A T 12 3 1 1 Text To Speech synthesis MO a LAST aA A aAa TRES a Tn 13 3 1 2 Exception Dictionary Read Mode ccceececenneeeeeeneeeeeeeaaeeeeeeaaeeeeeeaaeeeeeeaaeeeeeeaaaeeeesaaeeeeeeaaaaees 14 3 1 3 Hardware sound output busy signal sse ee eee eee 15 32 Control Codes Specihnicaiong sse 17 3 2 1 beveli C ntrol Code TTT 17 3 2 2 R a oeoo e css aa sce E A ade ee ee E E As es 18 3 2 3 Level ICONOS yrii ace aise a eat deed nes A ban Aa aa Anant aa aaa as a can dads 20 3 2 4 Commaid EE ee iea enea E A e E 22 4 Rules to be applied sisina an nanena raaraa a aaaea ag Ea SAS SE Seas s Seas aca celia 23 dlls BENIN ee eee ee ey rere morse reer ere ey A rote arene eee c
14. SCII codes 4 4 Dash The presence of a dash between two words is used by the system to recognise a hyphenated word or to apply liaisons between the two words The presence of a dash between two digits is used to recognise a scientific expression The correct use of the dash is therefore very important 4 4 1 between words e Hyphen When the dash is directly connected to the first word and just before a carriage return it is used to apply a liaison between the two words Example Between lines demons tracion will be pronounced demonstracion e Ignored When the dash is directly connected to the words it is ignored and translated like a space Example Ja n Andalusia will be pronounced Ja n Andalusia Dash between digits When the dash is directly connected to the first digit and between two digits it is pronounced gui n When the dash is preceded by a space character and directly connected to the second digit it is pronounced minus In all the other cases it is ignored and translated as a space character Example 34 35 will be pronounced treinta y cuatro gui n treinta y cinco 34 35 will be pronounced treinta y cuatro treinta y cinco 34 35 will be pronounced treinta y cuatro menos treinta y cinco The dash will be pronounced Minus with the control As 4 Depending on the platform gt See chapter Control code specification OKI SCP middle ware Spanish Text To Speech Users Manual
15. accepted 2 2 3 MSM7630 Start up Sequence MSM7630 operates under the following sequence when reset is applied Make reference to the flow chart when designing a text to speech synthesiser device that uses MSM7630 OKI SCP middle ware Spanish Text To Speech Users Manual Reset Applied Read Configuration Decide start program Initialize Memory SIO Driver TMR Read Configuration Status Check HF Initialize Open PIO Initialize Open SIO Micro controller 8bit serial POBF Z DSR CTS no PIBF Z Active yes Start DA Output Output DAO1 To TTS Main Program fig 2 1 ROM accesses are granted immediately after reset A 23 1 will fluctuate at this time Cache reads are performed so in particular the three low order bits will continuously change Active signals at this time will be as follows A 23 1 especially A 3 1 ROM RD Next the configuration register value will be read and the DRAM used will be set This starts DRAM refresh so the following signals will become active RAS CASO CAS1 Next the SIO drive will be initialized For male phoneme simplex data the mode will be set OKI SCP middle ware Spanish Text To Speech Users Manual the configuration register value will be read again and the interface used will be set Based on these settings the following signals will become active 8 Bit serial interface RTS TXD Micro controller interface POBF PIBF PD However these signals might not be s
16. ample Beginning of file Charles De Gaulle lt charl de gol gt ELAN informatique lt elan inkormatik gt i french company 14 Depending on the platform OKI SCP middle ware Spanish Text To Speech Users Manual End of file Note 5 2 Abbreviations lexicon After modifications the exceptions lexicon file must be reloaded in the memory 5 2 1 Using the lexicon If the abbreviation is written in the left column of the file it will be translated as indicated in the right column The translation writing of abbreviations uses a pseudo orthographic method For example the translation of the abbreviation U S A can be written United States of America 5 2 2 Adding an entry to the lexicon file With a text editor the user can add a entry to the abbreviation lexicon The abbreviation lexicon is a file called ABBREVIA RGS in the installation directory The maximum length of the lexicon depends on the available RAM resources 15 An abbreviation and its translation must be written on one line less than 256 characters long It is not necessary to respect the alphabetic order Finally the look up words are case sensitive Key characters list The characters indicate a comment which stop at the end of the line The space character or the tabulation separates the abbreviation field from the field of its translation The character indicates a word boundary in the abbreviation translation field Example
17. asic rules for spanish pronunciation can be stored in this user lexicon It contains a list of exception words with their corresponding pronunciation The pronunciation writing uses a pseudo orthographic method The pseudo orthographic method consists of writing the pronunciation with spanish alphabetical codes For example the pronunciation in spanish of the french word De Gaulle can be written lt de gol gt 5 1 2 Adding an entry to the lexicon file With a text editor you can add a new entry to the file called USERHISP EXC in the installation directory The maximum lenght of this file depends on RAM resources 11 Each exception must be written on only one line maximum 256 characters One exception can consist of one word or several consecutive words maximum 5 words It is necessary to put the same number of pronunciation words than of exception words Using punctuation marks in an exception is not forbidden Therefore it is impossible to write abbreviations in this file It is not necessary to respect the alphabetic order The look up words are case sensitive But if you add the option i the look up words are not case sensitive Key characters list The character indicates the end of the exception The codes between lt and gt indicate orthographic codes The two characters indicate word boundaries The two characters indicate comments The two characters i are optional and indicate to ignore case Ex
18. cept in exception dictionary read mode These codes codes primarily control speech quality Commands Control codes Valid in text to speech synthesis mode Commands control the speech synthesis sequence OKI SCP middle ware Spanish Text To Speech Users Manual 3 1 1 Text To Speech synthesis mode IN this mode sentences are input and then speech synthesised MSM7630 detects a termination in the input text by a termination character and starts the speech synthesizing operation Returning synthesis termination code HOST SCP Specify synthesis termination codes to be returned Speech synthesis Synthesis termination code Speech synthesis Synthesis termination code Specify no synthesis termination codes to be returned Speech synthesis fig 3 1 In the text to speech synthesis process MSM7630 normally just synthesizes speech from accepted test and does not return anything so a host cannot inspect MSM7630 software status For these case MSM7630 can be made to return a synthesis termination code each time synthesis processing of s sentence completes each time the synthesized sound is output by specifying that a synthesis termination code is to be returned refer to Control Codes Commands 1 Level 1 When a synthesis termination code has been specified to be returned only the response request code D 04H not the termination characters will be recognized as a terminator The host appends the response request code D
19. een as active for data Finally initialization of DA register internal values will begin and DAO1 pin output voltage will become active 1 5Volt Control will then jump to the main routine After this the individual interface will wait for input The above start up sequence needs about 700mSec MSM7630 does not perform self diagnostic as part of its start up process OKI SCP middle ware Spanish Text To Speech Users Manual 3 Text To Speech program specification 3 1 Operating Mode MSM7630 has the operating modes shown in the table below The operating mode is selected by an operating mode specification refer to the control code command listing in Appendix Table The default mode is text to speech synthesis mode When in this mode input sentences can be output as synthesized speech table 3 1 Function Text To Speech synthesis mode Unused Exception dictionary read mode Control codes and commands are provided to control MSM7630 operation The validity of control codes and commands differs depending on the operating mode The table below gives a summary of control codes and commands table 3 2 Category Function Level1 control Escape codes Valid except in exception dictionary read mode These codes codes primarily set the initial operating state of MSM7630 Level2 control Text related Valid in text to speech synthesis mode These code primarily codes control how sentences are read Level3 control Text related Valid ex
20. mples 16 03 1994 16 03 1994 16 3 1994 and 16 3 1994 will be pronounced el dieciseis de tres de mil novecientos noventa y cuatro 16 03 94 16 03 94 16 3 94 and 16 3 94 will be pronounced el dieciseis de tres de noventa y cuatro 16 03 and 16 03 will be pronounced el dieciseis de tres 45 9 1989 will not be processed as a date because 45 gt 31 and will be pronounced cuatro cinco pause nueve pause un nueve ocho nueve Note It is possible to pronounce 16 03 like a phone number using the control t 4 7 5 Currency Examples 5 13 ptas and 5 13ptas will be pronounced cinco pesetas trece 5 1 ptas will be pronounced cinco pesetas diez 5 56 FF will be pronounced cinco francos frances cicuenta y seis 4 7 6 Telephone numbers Examples 535 39 35 will be pronounced quinientos treinta y cinco pause treinta y nueve pause treinta y cinco 91 535 39 35 and 91 535 39 35 will be pronounced noventa y uno pause quinientos treinta y cinco pause treinta y nueve pause treinta y cinco 4 7 7 Scientific expressions Examples 02 123454 will be pronounced cero dos pause un dos tres cuatro cinco cuatro 12 2345 456 will be pronounced un dos pause dos tres cuatro cinco cuatro cinco seis ab12x will be pronounced aa bee doce equis When there is a combination of digits and letters the system pronounce the letters letter by letter and the numbers by group of more 3 digits
21. n tere ai as 23 4 1 1 Number of characters ascites ive tees ee i ee nn ee aie eed 23 4 1 2 Number of Words c 0 6hee ee A bee aad de et cde ete eee 23 BEI o HTHH T 23 Ae TENA q o A 24 44 Das A T 24 4 4 1 A AO 24 4 5 RUNNER 25 4 5 1 List of pronunciations recognised by the system and their effectS oooooccccnonnccccnnncccnononaconononncinns 25 4 5 2 Automatic breaks a die didas 25 4 5 3 IS TTT TTT 25 4 6 Acronyms and abbr viationS in a ad 25 4 6 1 List of acronyms and abbreviations of the Savate ee e e e 25 4 6 2 List of acronyms and abbreviations Of the USer ee e e e 26 ALT AAA i area a a e ai aiaeei aiaeei ists 26 4 7 1 A anan eA E Eaa A Ara a aaa Aaa saan Aa ceed Aaa aa aa a aAa 26 4 7 2 UI E EE E E IAE TETT TET 27 4 7 3 STH ss TTT 27 4 7 4 ST T TT TTT TTT 27 4 7 5 E TI eT 27 a OKI SCP middle ware Spanish Text To Speech Users Manual 4 7 6 Telephone numbers iii A ete chive S A 27 4 7 7 Scientific expressin Sns dpi wares eee eel ade Sel tae vied eileen 27 5 User lexICONS iii ais 29 51 Exceptions TICO Ma det dos 29 5 1 1 Usingithe SiC eii adorada 29 5 1 2 Adding an entry to the lexicon file eee eee ee eee 29 5 2 Abbreviations TEX ICOM dns isn na hedh Your cease ada Voie da avin e 30 5 2 1 DA T 30 5 2 2 Adding an entry to the lexicon file cece eee ORE aa 43 E E E RAT 30 6 APPENDIX A List of ASCII codes translated eeeeeeeeeeeeeeeeeeeeeeeesesesesesesesesesesese 31 6
22. peech Users Manual ignored ignored ignored sign micro ignored ignored ignored sign ohm ignored ignored ignored ignored ignored ignored n small letter ignored ignored ignored ignored ignored ignored ignored sign grad ignored ignored ignored ignored sign cuadrado c e i n ignored ignored
23. s trecientos cuarenta y tres 567 123 78 890 556 will not be processed as an integer because the groups separated by are not composed of 3 digits It will be pronounced digit per digit with a pause at the point 012 will be pronounced cero un dos e Decimal numbers They are correct if there is no space character between the and the numbers for instance 36 55 is correct but 36 55 is not Examples 4 56 will be pronounced cuatro coma cincuenta y seis 3 4 will be pronounced menos tres coma cuatro 0 456 will be pronounced cero coma cuatrocientos cinquenta y seis 1 234 456 123 will be pronounced 1 million 234 mil 456 coma ciento veintitres 1912 123 will be pronounced mil novecientos doce pause ciento veintitres e Ordinal numbers An ordinal number is a number terminated by 2 or by Examples 20 will be pronounced vig sima 20 will be pronounced vig simo 8 The point can be used to separate groups of 3 digits in large numbers OKI SCP middle ware Spanish Text To Speech Users Manual 4 7 2 Time Examples 5h and las 5h will be pronounced as cinco 5 45 5h45 and las 5h45 will be pronounced las cinco cuarenta y cinco 5 h will be pronounced cinco hora 4 7 3 Duration Examples 5h45mn and 5h45m will be pronounced cinco horas cuarenta y cinco minutos 4 7 4 Date A date format is as follow three or two numbers separated by points or slashes Exa
24. speech synthesis parameters MSM7630 will then return to an input wait state 3 2 4 2 Initialize Stops processing of the current operating mode Returns all Level 1 to 3 Code settings including mode specification to their defaults table 3 1 AR 12H Stop processing of the current operating mode OKI SCP middle ware Spanish Text To Speech Users Manual 4 Rules to be applied 4 1 Sentence 4 1 1_ Number of characters A sentence must not be more than 900 characters long control codes included Longer sentences will be truncated between two words to produce two or several sentences which will be less than 900 characters long 4 1 2 Number of words A sentence must not be more than 70 words long control codes3 excluded lf a sentence contains more than 70 words without punctuation the system automatically inserts a full stop 4 2 Word A word must not be more than 64 characters long Longer words will be truncated to 64 characters to produce two or several words of less than 64 characters 1 See chapter Control code specification 2 Overflow may be caused by the translation of numbers and acronyms For example the number 033544628 which has 9 characters will have 46 characters after translation 3 See chapter Control code specification OKI SCP middle ware Spanish Text To Speech Users Manual 4 3 Character A character must be coded in IBM extended ASCII or in ISO 8859 14 Refer to appendix A for the translation of A
25. trol MSM7630 s speech synthesis operations before starting Some are sent alone and some are sent inserted anywhere between sentences or words in the text 3 2 1_Level1 Control Code Level 1 control codes are output before the text file to set the operating state of MSM7630 Text characters are specified in half size capitals to follow the escape code 1BH Lists the Level 1 control code table 3 1 Operating mode Synthesis termination code 3 2 1 1 Code format Specifies the code format of input text The word dos refers to IBM extended characters table 3 1 TTIm T o 3 3 2 1 2 Operating mode specification Specifies the MSM7630 s operating mode table 3 1 Code format 1 ESC MO Text to Speech synthesis mode default OKI SCP middle ware Spanish Text To Speech Users Manual ESCM ESCM ESC M3 Exception dictionary read mode 3 2 1 3 Synthesis termination codes returned not returned This feature specifies whether or not a synthesis termination code is to be returned after synthesis ends for each sentence Since MSM7630 normally speech synthesizes the text it receives without returning anything the host cannot inspect its status Therefore while the host shows text one character at a time on its display and sends the text to the MSM7630 for speech synthesis processing the display and synthesized sounds may not be synchronized since there is a process delay from text input to synthesis start S
26. ynthesis termination codes are used to synchronize the host and MSM7630 processes table 3 1 Code format 1 ESC EO Do not return synthesis termination codes default note 1 The terminating character will be recognized as the end of text If text analysis is not possible then the portion of text that cannot be analyzed will be skipped but the speech synthesis process will be performed 2 ESC E1 Return synthesis termination code Instead of a terminating character only the response request code D 04H will be recognized as the end of text note 2 ESC ED Return to default setting Note 1 fig 3 2 shows the format of synthesis termination codes Note 2 The response request code is appended after the text s terminating character 3 2 2 Level 2 Control Code Level 2 control codes not only set the operating state prior to sending a text but can also used between sentences in a text They are specified with characters and affect text following the control code table 3 1 Level 2 Control Code numeric form pronunciation These controls allow the numeric forms to be pronounced in several ways depending on the context The default mode is usual There are 6 control codes usual scientific commercial date telephone roman OKI SCP middle ware Spanish Text To Speech Users Manual table 3 2 Control Code INFORMATION VALUE u To restore the default mode No information value s To pronounce scientific
Download Pdf Manuals
Related Search