WTS701EF - Digi-Key
Contents
1. WYSIWYG whisywig THE the 26 WTS701EF USER S MANUAL 6 Winbond Electronics Corp 5 VERSION HISTORY VERSION DATE PAGE DESCRIPTION 1 00 June 2003 All Initial Version The contents of this document are provided only as a guide for the applications of Winbond products Winbond makes no representation or warranties with respect to the accuracy or completeness of the contents of this publication and reserves the right to discontinue or make changes to specifications and product descriptions at any time without notice No license whether express or implied to any intellectual property or other right of Winbond or others is granted by this publication Except as set forth in Winbond s Standard Terms and Conditions of Sale Winbond assumes no liability whatsoever and disclaims any express or implied warranty of merchantability fitness for a particular purpose or infringement of any Intellectual property Winbond products are not designed intended authorized or warranted for use as components in systems or equipments intended for surgical implantation atomic energy control instruments airplane or spaceship instruments transportation instruments traffic signal instruments combustion control instruments or for other applications intended to support or sustain life Further Winbond products are not intended for applications wherein failure of Winbond products could result or lead to a situation where2. Winbond Electronics Corp WTS701EF ENGLISH FEMALE VERSION TEXT TO SPEECH USER S MANUAL Publication Release Date June 2003 1 Revision 1 00 WTS701EF USER S MANUAL A Winbond Electronics Corp TABLE OF CONTENTS I noon I 4 2 Winbond Text To Speech Technology Overview 5 2 1 Overview of Device Initialization Sequence 5 2 2 Description of User Interface s s s ee E E see ener enne nennen nennen ER 5 2 3 Description of WTS701 State 6 2 3 1 Text To Speech Mechanism iir ree eret res ep dae aue aer eode eeu ee 6 2 3 2 Text NorttialiZatiob iet rr RR TR EET PO 7 2 3 3 Letter to Phoneme 510 ee eee 7 2 3 4 lt Phoneme dad d A A AR 7 24 SPI Commands Overview 8 2 4 1 Command e d aa am At 8 2 5 Description of Phonetic Alphabet Feature sees sees eee eee 9 2 6 Description of Abbreviations eene eene 12 3 Rules to be Applied arree eeen eoe toe edd eese 13 Del POX mE 13 2 7 etae 13 3 36 Word 13 14
3. 0 to 7 OdB to 28dB The default volume level is 3 12dB Any number that is greater than 7 will be set to 0 A space character is required after the level number Publication Release Date June 2003 15 Revision 1 00 WTS701EF USER S MANUAL 6 Winbond Electronics Corp The following example shows how these control characters can be used together handling not only capitalization and pronunciation issues but also establishing emphasis and conveying some emotion V3 Well V 754 to tell you the truth V S2 it was Q2 V JOHN U0 V who V S wrote the report 5 V not Q1 V Pg0iylerm0o V 3 6 DASH e When a dash appears between words with no space preceding or following it is recognized as hyphen in a hyphenated word There is no abbreviation support for the hyphenated words Example three year old will be pronounced as three year old e When there is a space preceding or following the dash it will be pronounced as dash Example three year will be pronounced as three dash year three year will be pronounced as three dash year e When used in front for a digit it will be pronounced as minus Example 5 will be pronounced as minus five 3 7 SLASH When a slash appears between words with no space preceding or following it is treated as a space Otherwise it will be pronounced out as slash There is no
4. 3 5 Control 14 16 SNL MEE 16 SA NR m 16 3 9 ntemet E mail Address ent rere e AE E 17 Im 17 3 11 Abbreviat ions uc a anan Ra R a R R 18 3 12 Numeration U 18 EB DS 18 S Ti S 18 EM PAS E RU 19 3 12 4 Dollar SiO taedet ee eese e oed D d 19 3 12 5 Combination of digits and other characters 19 EN PAD Sun s 19 WTS701EF USER S MANUAL QWinbond Electronics Corp 4 urbe came 20 4 1 Input Character Table etse eto nee O EE EE E 20 4 2 Default Abbreviation List sss s s s 21 5 Version HiSfOTY 27 Publication Release Date June 2003 3 Revision 1 00 WTS701EF USER S MANUAL 6 Winbond Electronics Corp 1 INTRODUCTION The WTS701 is a high quality fully integrated single chip Text to Speech solution that is ideal for use in applications such as automotive appliances GPS navigation systems cellular phones and other porta
5. abbreviation support in this case Example boy girl will be pronounced as boy girl boy girl will be pronounced as boy slash girl 3 8 DOT When is used within a decimal number it will be pronounced as point Example 16 WTS701EF USER S MANUAL 6 Winbond Electronics Corp 2 6 will be pronounced as two point six When is used at the end of a word or a sentence it is treated as silence please see section 3 7 for more details For all the other cases that are not described above will be pronounced out as dot 3 9 INTERNET E MAIL ADDRESS Any string which contains www com or gov as a substring will be treated as a web or e mail address Example vvinbond com will be pronounced as winbond dot com www irs gov will be pronounced as vv vv w dot ir s dot gov 3 10 PUNCTUATION Punctuation Pause Duration 0 2 0 6 0 5 sec 0 5 sec 0 5 sec Note A space character is required at the end of ellipse While pauses are insterted for punctuation punctuation also influences the pronunciation of neighboring words Punctuation should not added in violation of standard English usage in order to insert or extend pauses this should be done instead with the pause control flag control Q 0x11 see section 3 5 In accordance with standard Engl
6. configuration registers for the users requirements Table 1 Initialization Command Sequence State Command Description POWER State after power on or RST command DOWN SCLC Set clock configuration PWUP Power up device IDLE SCOM Set up communication register to enable interrupts SCOD Set up CODEC configuration if used SAUD Set up audio control register SVOL Set the initial volume level SSPD Set the initial speech output speed level SPTC Set the initial speech pitch level 2 2 DESCRIPTION OF USER INTERFACE As a real System On Chip solution the WTS701 performs the overall control functions for host controller and text to speech processing The WTS701 system architecture consists of the following functions e Serial interface to monitor the SPI port and interpret commands and data e Text normalization module to pre process incoming text into pronounceable words e Words to phoneme translator which converts incoming text to phoneme codes e Phoneme mapping module that maps incoming phonemes to words sub words syllables or phonemes present in the MLS memory e Volume and speed adjustments Digital and analog output blocks for off chip usage Publication Release Date June 2003 5 Revision 1 00 WTS701EF USER S MANUAL Electronics Corp The WTS701 system performs text to speech synthesis based on concatenative samples The units for concatenation can vary
7. from whole words down to phoneme units The convention is that the larger the sub word unit used for synthesis the higher the quality of the speech output A corpus of pre recorded words is stored in Winbond s patented multilevel storage MLS memory and a mapping of the various sub word parts is held in a lookup table The speech creation is achieved by concatenation of these speech elements to produce words The system process flow is shown in Figure 1 WTS701 Serial Text symbols amp Control Text Normalization Letter to Phoneme Phoneme Mapper Digital output MLS Memory Speech Figure 1 WTS701 System Process Flow 2 3 DESCRIPTION OF WTS701 STATE MACHINE 2 3 1 Text To Speech Mechanism The text to speech component of the system consists of three principal blocks Text normalization Letter to phoneme conversion e Phoneme mapping WTS701EF USER S MANUAL 6 Winbond Electronics Corp 2 3 2 Text Normalization Text normalization involves the translation of incoming text into pronounceable words It includes such functions as expanding abbreviations and translating numeric strings to spoken words It involves a certain amount of context processing to determine correct spoken form In addition the WTS701 looks into the abbreviation list stored in the device s internal memory and converts acronyms abbreviations or special characters such as Instant Messaging icons or emoti
8. specify the pronunciation of unusual words outside the central core of English words covered by the letter to phoneme module For example the following transcriptions can be provided for French President Jacques Chirac s name PZlak PSOirlak Tweaking pronunciations Publication Release Date June 2003 11 Revision 1 00 WTS701EF USER S MANUAL 6 Winbond Electronics Corp The phonetic alphabet can also be used to tweak pronunciations Many English words even common words have multiple valid transcriptions but in some cases the output for each is not equally as good For example the default transcription for current for the VVTS7O1EF is PkIRrOInt The transcription Pk1ROInt is equally valid but the output is much worse Exploring variant phonetic transcriptions can solve many output problems Developers of TTS applications are often tempted to fix pronunciation problems by using abberant spellings but they should use transcriptions instead as phonetic transcriptions provide much greater control Indeed many issues can only be addressed by providing transcriptions 2 6 DESCRIPTION OF ABBREVIATIONS FEATURE The WTS701 has support for entering and using custom abbreviations in addition to the general abbreviation table supported internally by the WTS701 These supplement the default abbreviation support See sections 3 8 and 4 2 There are 2K bytes of flash memory reserved for this purpose After the WTS70
9. 1 internal software has been initially programmed this entire area is free and available for custom abbreviations The commands associated with custom abbreviations are Command Command Command Byte Data Byte ABBR_ADD Oxaf 0x00 Adds a new abbreviation to the abbreviation table abbreviation in the WTS701 See below for the format of the data abbreviation data ABBR DEL 0x83 0x00 Deletes an existing abbreviation from the abbreviation abbreviation table in the WTS701 See below for data the format of the abbreviation data ABBR NUM Oxc8 0x00 0x00 Returns the number of abbreviation currently active 0x00 in the abbreviation table of the WTS701 ABBR MEM Oxc7 0x00 0x00 Returns the number of free bytes in the 0x00 abbreviation table of the WTS701 ABBR RD 9 0x00 2048 Returns the abbreviation table contents from the 0x00s WTS701 See below for the format of the abbreviation table data previously stored into xdata after this command ENTER RRSM 0x0c 0x00 Causes the xdata and code store memory to swap spaces The WTS701 begins to execute code For both ABBR_ADD and ABBR_DEL the abbreviation data should be formatted as follows input_abbreviation comma output_string semicolon For example the following associates the abbreviation Fr with Father Fr Father If this data were entered with ABBR ADD Fr Miller would be read Father Mille
10. BYTE 0 STATUS BYTE 1 DATAO DATA DATAn p time Figure 2 SPI Transaction Format 2 4 1 Command Classes The SPI transactions to the WTS701 fall into four classes The four classes represent variations in how the command and any associated data is handled The class of a command is defined by the two most significant bits of the command byte A summary of the command classes is given below CLASS 0 COMMANDS These are commands that are executed irrespective of the state of the WTS701 That is the command will execute even if the device is busy or powered down These commands are executed internally by a hardware command interpreter All commands not of class 0 require that the WTS701 be in a powered up state Example of class 0 command is the Read Status RDST command CLASS 1 COMMANDS Class 1 commands require interpretation by the internal firmware of the WTS701 Class 1 commands consist only of a command byte and command data byte Any further data sent in a transaction is ignored Class 1 commands are most often used for setting a configuration register in the device or sending commands that have no data such as the conversion pause PAUS command WTS701EF USER S MANUAL 6 Winbond Electronics Corp CLASS 2 COMMANDS Class 2 commands have associated data After the command word any data bytes following are loaded into an internal FIFO buffer for processing If this FIFO becomes full the R B signal is as
11. ble products or accessories The WTS701 product accepts ASCII Unicode and Big5 for Mandarin input via a SPI port and converts it to spoken audio via an analog output or digital CODEC output The VVTS701 integrates a text processor smoothing filter and multi level memory storage array on a single chip Text to speech conversion is achieved by processing the incoming text into a phonetic representation that is then mapped to a corpus of naturally spoken word parts The synthesis algorithm attempts to use the largest possible word unit in the appropriate context to maximize natural sounding speech quality The speech units are stored uncompressed in a multi level non volatile analog storage array to provide the highest sound quality to density trade off This unique single chip solution is made possible through Winbond s patented multilevel storage technology Voice and audio signals are stored directly into solid state memory in their natural uncompressed form providing superior quality voice reproduction The chip can be programmed through the SPI port allowing downloading of different languages and speaker databases when made available by Winbond WTS701EF USER S MANUAL Electronics Corp 2 WINBOND TEXT TO SPEECH TECHNOLOGY OVERVIEW 2 1 OVERVIEW OF DEVICE INITIALIZATION SEQUENCE Configuration After power on or a Reset command RST the WTS701 processor can be configured for operation This involves initializing the internal
12. cons into the appropriate text representation The default abbreviation list supported by the WTS701 is a general one that cannot be modified by the user to match the domain that the text is being loaded from But the default list can be overridden by the user abbreviation list This enables a flexibility of adding abbreviation specifically for the text either by the developer or even the end user to best customize the product for its preferences Instant Messaging or Short Messages Service SMS unique characters are supported through this functionality as well defining the icon ASCII Unicode Big5 text and its replacement The default abbreviation list supported is described in the specific language release letter 2 3 3 Letter to Phoneme Conversion Once the data stream has been translated to pronounceable words the system next determines how to pronounce them This function is obviously highly language dependent For a language such as English it is impossible to break this task down to a set of definitive rules The task is achieved by a combination of rule based processing together with exception processing 2 3 4 Phoneme Mapping This algorithm maps phoneme strings into the MLS phonetic inventory This task falls into two portions First the word must be split into sub word portions This splitting must be done at appropriate phonetic boundaries to achieve high quality concatenation Once a sub word unit is determined the inventory is
13. d as one point two three e Ordinal numbers Any number terminated by 2 3 4 5 6 7 8 9 0 th is recognized as an ordinal number Example 217 will be pronounced as twenty first 3 12 2 Time The time representation format is 00 00AM or 00 00PM one semicolon between 2 integers AM am or PM pm is optional but if needed it must be added right after the second integer Example 9 30pm will be pronounced as nine thirty m 18 WTS701EF USER S MANUAL 6 Winbond Electronics Corp 3 12 3 Date The support for dates goes from year 2000 to 2099 and the format is MM DD YY For the dates before year 2000 please use MM DD YYYY Example 1 15 03 will be pronounced as one fifteen two thousand three 3 12 4 Dollar sign The combination of a and any integer or any decimal number is recognized as money No space is allowed in between Example 15 99 will be pronounced as fifteen dollars and ninety nine cents 500 million gt five hundred million dollars 3 12 5 Combination of digits and other characters This occasion happens mostly in telephone numbers Example 408 123 4567 will be pronounced as four zero eight one two three four five six seven A pause will be placed after 8 and another pause will be placed after 3 123 4567 will be pronounced as one two three four five six seven 3 12 6 Percent S
14. ed 0 14 Reserved 0x3d 0 54 close bracket 0 74 Undefined Oxle Reserved Ox3e Undefined OxSe A Ox7e Undefined 0 1 Reserved Ox3f OxSf underscore Ox7f Undefined 20 WTS701EF USER S MANUAL 6 Winbond Electronics Corp Note 1 Characters ranges from 0x09 to 0 00 escape characters will be treated as white space Note 2 The characters marked Reserved above are the characters reserved for future system development Unpredicted behavior may occur if these characters are used 4 2 DEFAULT ABBREVIATION LIST The list below is stored in WTS701 and used for all incoming text Note also that there is no abbreviation support in both U1 and U2 modes Sen Senator Rep Representative Jan January Feb February Apr April Jun June Jul July Aug August Sep September Sept September Oct October Nov November Dec December Mon monday Tue tuesday Wed wednesday Thu thursday Fri friday Sat saturday Sun sunday A M A M P M P a m T M M Publication Release Date June 2003 21 Revision 1 00 6 Winbond Electronics Corp pm P M PM P M PhD P Mr Mister Mrs Missus Ms Miz Dr Doctor Jr Junior Esq Esquire Pres President Prof Professor Sgt Sargean
15. he system will pronounce every all capital word as a normal word as long as the word is pronounceable In this mode no abbreviations will be detected A space character is required after 1 Example U1 ROM will be pronounced as rom whereas U0 ROM will be pronounced as r o m U2 Force every word to be spelled out regardless of case or length In this mode abbreviations will be detected A space character is required after 2 Example U2 GOOD will be pronounced as g o o 4 U2 good will be pronounced as g o o d U3 Force every all capitol word to be spelled out regardless of its string length unless the word is an abbreviation established in the system Abbreviations will be treated as indicated in the built in abbreviation list see 04 2 or by the user see 02 6 It simlar to U0 except for the insensitivity to string length A space character is required after 3 Example U3 HELLO will be pronounced as h ell 0 Control V 0x16 Volume control flag This command flag changes the volume V Increase the playback volume by 1 level A space character is required after Example V r Hello world V Decrease the playback volume by 1 level A space character is required after Example V Hello world AVX Set the playback volume to level X X is an integer and its valid range is from
16. ign Example 1279 will be pronounced as twelve percent Publication Release Date June 2003 19 Revision 1 00 WTS701EF USER S MANUAL Winbond Electronics Corp 4 APPENDICES 4 1 INPUT CHARACTER TABLE 0x0 Reserved 0x20 Space 0x40 0x60 Undefined 0 1 Reserved 0x21 0x41 A 0x61 a 0x2 Reserved 0x22 Undefined 0x42 B 0x62 b 0x3 Reserved 0x23 0x43 0 63 0 4 Reserved 0x24 0x44 D 0x64 d 0x5 Reserved 0x25 0x45 E 0x65 e 0x6 Reserved 0x26 amp 0x46 F 0x66 f 0x7 Reserved 0x27 apostrophe 0x47 G 0x67 g 0x8 Reserved 0x28 0x48 H 0x68 h 0 9 Space 0x29 0x49 I 0x69 I Oxa Space 2 0 4 J Ox6a j Oxb Space 0x2b Ox4b K 0x6b k Space 2 comma 4 L Ox6c 1 0 Space 0 24 dash 0 44 0 64 m Oxe Reserved 2 period 4 N n Oxf Reserved Ox2f slash Ox4f o Ox6f o 0x10 P 0x30 0 0x50 0 70 0 11 Q 0x31 1 0x51 Q 0x71 q 0x12 Reserved 0x32 2 0x52 R 0x72 r 0x13 S 0x33 3 0x53 S 0x73 s 0x14 Reserved 0x34 4 0x54 T 0x74 t 0x15 U 0x35 5 0x55 U 0x75 u 0x16 V 0x36 6 0x56 V 0x76 M Ox17 Reserved 0x37 7 0x57 W 0x77 W 0x18 Reserved 0x38 8 0x58 X 0x78 x 0x19 Reserved 0x39 9 0x59 Y 0x79 y Oxla EOT 0x3a colon 0x5a Z Ox7a 7 Ox1b Reserved 0x3b Undefined Ox5b open bracket Ox7b Undefined Oxlc Reserved 3 Undefined 5 back slash Ox7c Undefin
17. in personal injury death or severe property or environmental injury could occur Winbond Electronics Corp Headquarters No 4 Creation Rd 111 Science Based Industrial Park Hsinchu Taiwan TEL 886 3 5770066 FAX 886 3 5665577 http www winbond com tw Taipei Office 9F No 480 Pueiguang Rd Neihu District Taipei 114 Taiwan TEL 886 2 81777168 FAX 886 2 87153579 Winbond Electronics Corporation America 2727 North First Street San Jose CA 95134 U S A TEL 1 408 9436666 FAX 1 408 5441797 http Awww winbond usa com Winbond Electronics Corporation Japan 7F Daini ueno BLDG 3 7 18 Shinyokohama Kohokuku Yokohama 222 0033 TEL 81 45 4781881 FAX 81 45 4781800 Winbond Electronics Shanghai Ltd 27F 299 Yan An W Rd Shanghai 200336 China TEL 86 21 62365999 FAX 86 21 62356998 Winbond Electronics H K Ltd Unit 9 15 22F Millennium City No 378 Kwun Tong Rd Kowloon Hong Kong TEL 852 27513100 FAX 852 27552064 Please note that all data and specifications are subject to change without notice lication Release Date June 2003 All the trademarks of products and companies mentioned in this datasheet belong to their respective owners Revision 1 00 This product incorporates SuperFlash technology licensed from SST 7
18. irginia VVT VVisconsin VVY VVyoming N North S South E East Ww West LA Los Angeles SF San Francisco S San Jose Ans Answer Q Question Asst Assistant Atty Attorney Bldg Building asap ASAP cm centimeters mm millimeters ft feet yd yards yds yards 24 WTS701EF USER S MANUAL QWinbond Electronics Corp pt pints pts pints qt quarts qts quarts km kilometers so mi miles deg degrees cu cubic diam diameter FAX fax r Geog geography govt government woe min minute mins minutes hr hour hrs hours VIS doz dozen years Inc incorporated KVV kilovvatt Ib pounds Ibs pounds mngr manager Nat National NASDAQ nasdaq ok OK w oz ounces PI place rcvd received rpm RPM tel telephone ty TV Publication Release Date June 2003 25 Revision 1 00 WTS701EF USER S MANUAL QWinbond Electronics Corp woe ys versus U S yoo ess U S A yoo ess ay VISA veeza w with wrt with respect to wk week Xmas Christmas etc etcetra c o care of w o without IIT the third
19. ish orthography all of these punctuation markers should immediately follow the preceding word with no internening space Violating this norm can adversely affect output quality For example there should be no space before the period in Believe it or not If Believe it or not is entered instead the output for not will not be as appropriate to the context The only Publication Release Date June 2003 17 Revision 1 00 WTS701EF USER S MANUAL 6 Winbond Electronics Corp exception is for command character sequences e g Good night V my sweet PprlIns see section 3 5 3 11 ABBREVIATIONS The system recognizes a default set of frequently used abbreviations see section 5 2 and also allows users to add abbreviations of their own see section 2 6 Matching is case sensitive but both the presence and the absence of final periods are ignored in matching For example the abbreviation Feb is provided for February Both Feb and Feb would constitute matches but both feb and feb would not Note also that there is no abbreviation support in both U1 and U2 modes See section 3 5 3 12 NUMERATION 3 12 1 Numbers e Integers Example 10 will be pronounced as ten 10 000 will be pronounced as ten thousand 94087 will be pronounced as nine four zero eight seven e Decimal numbers Example 1 23 will be pronounce
20. n in the San Francisco Bay area and the transcriptions generated are consistent with this dialect Transcriptions can be modified to emulate some properties of other dialects including Standard British and Australian dialects but a complete transformation cannot be successful as many other dialect dependent features cannot be modified with transcriptions Abstractness These phonetic transcriptions are abstract representations and some transformations are performed on these transcriptions on the way to creating audio output Three of phonemes in the preceding table are commonly products of these transformations The vowel 0X is associated with both OI and Ox It can be used in input transcriptions but the user can use OI and Ox instead as selecting among the two as appropriate For example roses may also be transcribed as PrlozOIz The output will be identical Both q and P are associated with t Written can be transcribed as Pr1ItO0xn as well as PrlIqOxn and eighty may be transcribed as Plet0i as well as PlePOi The output will be identical in both pairs The user may simply use t except to force output with q or P Uses Greater control The phonetic alphabet feature may be used to specify particular pronunciations Examples Do you say PtOxml eP0o or PtOxmlat0o People from Missouri say they re from Pm lzlur x Foreign and unusual words The phonetic alphabet can also be used to
21. peed by level A space character is required after Example ASE Hello world S Decrease the playback speed by 1 level A space character is required after Example S Hello world SX Set the playback speed to level X X is an integer and its valid range is from 0 to 4 from fast to slow The default speed level is 2 Any number that is greater than 4 will be set to 4 A space character is required after the level number Example S0 Hello world 14 WTS701EF USER S MANUAL 6 Winbond Electronics Corp Control U 0x15 All capital word control flag This command flag tells the system whether a string comprised exclusively of capital letters should be spelled out or not U0 Force all short all capital words 4 or less letters to be spelled out unless the word is an abbreviation established in the system Abbreviations will be treated as indicated in the built in abbreviation list see 4 2 or by the user see 2 6 Longer all capital words 5 or more letters will be pronounced if deemed to be a word pronounceable in English This is the default setting A space character is required after 0 Example UO USA will be pronounced as us a U0 CA will be pronounced as California A built in abbreviation ZZZZZZZ will be spelled out because it is not pronounceable in English UI Avoid spelling out words in all capitals T
22. r See section 3 8 for 6 Winbond Electronics Corp WTS701EF USER S MANUAL constraints on abbreviation matching An abbreviation can also map onto multiple words e g BC before Christ or BC British Columbia This feature can also be used to provide transcriptions according to the conventions in section 2 5 For example NASA Pn1 s0x would provide the common pronunciation of NASA This can also be done for ordinary spelled out words as a means for consistently providing user transcriptions Thus if Chirac PSOirlak were added the pronunciation for the input text Chirac would consistently be PSOirlak The abbreviation table is provided in the following format when an ADDR_RD command is issued abbreviation number input abbreviation output string Abbreviations are listed and numbered in the order in which they are added If an abbreviation is deleted the text deleted appears after the input abbreviation Example 01 Fr Father 02 BC deleted before Christ 03 BC British Columbia 04 NASA Pn1 s0x 05 Chirac PSOirlak 3 RULES TO BE APPLIED 3 1 TEXT There is no pre set character length limit for an input text 3 2 SENTENCE There is no pre set character length limit for an input sentence 3 3 WORD TS Publication Release Date June 2003 Revision 1 00 WTS701EF USER S MANUAL 6 Winbond Electronics Corp Any space bounded string is
23. searched to determine if a match is present A matching weight is assigned to each match depending on how closely the phonetic context matches Each sub word has a left and right side context to match as well as the phoneme string itself If no suitable match is found in the inventory then the sub word is further split in a tree like manner until a match is found The splitting tree is processed from left to right and each time a successful match occurs the address and duration of the match in the corpus is placed in a queue of phonetic parts to be played out the audio interface Publication Release Date June 2003 7 Revision 1 00 WTS701EF USER S MANUAL 6 Winbond Electronics Corp 2 4 SPI COMMANDS OVERVIEW The WTS701 is controlled by a series of SPI transactions to send commands to the device The general format of an SPI transaction is shown in Figure 2 A transaction is always started by sending a command word The command word consists of a command byte followed by a command data byte At the same time the status register is shifted out on the MISO line What follows depends on what command 15 sent The general case is that following the command word up to n bytes of data can be sent to the device and n bytes can be read from the device An SPI transaction is finished when SS is returned to the HIGH condition MSB LSB MOS CMD BYTE CMD DATA DATAO DATAT DATAn MISO 4 STATUS
24. serted LOW indicating that the host must pause data transfer An alternative to monitoring the R B line the R B bit of the status register can be monitored instead see subsection 7 3 2 or via the RDST command CLASS 3 COMMANDS Class 3 commands have data to return to the host The R B line will go to busy immediately following the command word indicating that the WTS701 is fetching the requested data Data is put into the BCNTO and 1 see subsection 7 3 4 registers and is read out in the two subsequent bytes after R B is released If more than two bytes are returned from the command R B will again be asserted until data is ready to read The primary Class 3 commands are to read the contents of internal configuration registers such as RREG command 2 5 DESCRIPTION OF PHONETIC ALPHABET FEATURE As indicated in 2 3 3 the VVTS701 covers spelled out pronounceable words into phonetic transcriptions ie a string of phonemes providing an abstract representation of the target pronunciation This feature allows the input to contain phonetic transcriptions instead of ordinary English text explicitly indicating the desired pronunciation Phonetic transcriptions can be sent directly to the WTS701 This can be done by embedding phoneme strings in the text stream for conversion To embed a phoneme string the string must be preceded by a control P P ASCII 0x10 character and terminated by a space character If there is following punctua
25. t Lt Lieutenant Lieut Lieutenant MPH mile per hour Col Colonel Gen General Sr Senior St Street Av Avenue Ave Avenue Rd Road Rt Route Rte Route Blvd Boulevard Terr Terrace w Exwy Expressway Expwy Expressway Drwy Driveway Dept Department Ct Court AL Alabama WTS701EF USER S MANUAL WTS701EF USER S MANUAL 6 Winbond Electronics Corp AK Alaska AZ Arizona AR Arkansas CA California Cal California Calif California Cal California DE Delavvare FL Florida Fla Florida GA Georgia IEEE T triple E ID Idaho IL Ilinois IN Indiana IA Iowa KS Kansas KY Kentucky IL Illinois MA Massachusetts MI Michigan MS Mississippi MN Minnesota MO Missouri MT Montana NV Nevada NH New Hampshire NM New Mexico NY New York NJ New Jersey NC North Carolina ND North Dakota OH Ohio Publication Release Date June 2003 23 Revision 1 00 WTS701EF USER S MANUAL 6 Winbond Electronics Corp PA Pennsylvania RI Rhode Island SC South Carolina SD South Dakota TN Tennessee TX Texas Tex Texas UT Utah VT Vermont VA Virginia VVA VVashington VVV VVest V
26. tion or word a space must still intervene For example The quick brown Pf1Aks jumped over the lazy Pdlcg The following table lists the phoneme symbols acceptable by the WTS701 English Female software version As the acceptable phoneme symbols are language and version dependent please refer to the specific language User s Guide for details regarding characters accepted and other development considerations Publication Release Date June 2003 9 Revision 1 00 WTS701EF USER S MANUAL QWinbond Electronics Corp Table2 Acceptable Phoneme Symbols Phoneme Example ls T x about x Note that each phoneme is represented by exactly one character and there must be an indication of stress before each vowel The digit 1 is used to indicate primary stress and each word has only one primary stress All other vowels are marked with the digit 0 10 WTS701EF USER S MANUAL QWinbond Electronics Corp Examples Input Phonetic transcription hi 1 phoneme followed by a 1 stress vowel phoneme Y cute kylut backpack b1 kp0 k encyclopedia 01150 1 1140 0 Constraints on stress All vowels may bear 0 stress However neither x nor X can bear primary stress marked with 1 The best primary stress alternative for these two is 1A but 11 may be more appropriate for X in some cases Dialect note The VVTS701EF presents a dialect of Standard American English spoke
27. treated as a word The maximum character length of a word is 53 Words 54 or more characters in length will be truncated after the 53 character 3 4 CHARACTER A character must be encoded in ASCII The set of ASCII characters defined by the system ranges from 0x00 to 0x7A excluding 0x22 0x3B 0x3C Ox3E and 0x60 please refer to section 5 1 for more details All the undefined characters will be deleted prior to letter to phoneme conversion 3 5 CONTROL CHARACTERS To allow users to manipulate the system functions via the input text certain control characters are defined as command flags Actual HEX values must be sent to the chip These should not be confused with the sequence of followed by a capital letter Control P 0x10 Phoneme control flag This command flag indicates a phoneme string is to follow A space character is required after the end of the phoneme string Example PprOEzOIntleSOIn will be pronounced as presentation Control Q 0x11 Pause control flag Pauses with variant length can be added within a sentence by using the QX flag X is an integer which indicates the pause duration in tenths of a second A space character is required after the pause duration digit Example Q10 will add in a one second pause Control S 0x13 Speed control flag This command flag controls the speed change AS Increase the playback s
Download Pdf Manuals
Related Search