SpeechLib is a Microsoft BASIC Library routine developed with the Macintosh 68000 Development System. It was written using the guidelines in the Microsoft publication, "Microsoft BASIC for the Macintosh - Building Machine Language Libraries". SpeechLib is used in conjunction with MacinTalk, the software speech synthesis resource developed by Apple Computer.
These are the icons for MacinTalk and SpeechLib. Both of these resources must be accessible by the Microsoft BASIC Interpreter in order to use the speech synthesis capabilities of MacinTalk from within BASIC. (Only BASIC versions 2.0 and higher are supported by SpeechLib.)
USING SPEECHLIB:
The Library routine has to be initialized before it can be accessed. Therefore, the following line should appear in your program before any speech calls:
LIBRARY "[Volume:]SpeechLib"
the Volume name must be used if SpeechLib is on a disk other than the disk containing Microsoft BASIC.
If BASIC cannot free enough memory to initialize the library, an "illegal function call" error will be generated. Try changing the amount of memory allocated to the heap with the CLEAR command if this happens. (The CLEAR command is documented on page 109 of the BASIC manual, also check the memory management section beginning on page 71.)
Once BASIC has successfully opened the library, the SpeechLib routines can be used to access MacinTalk. SpeechLib 3.0 fully supports all the existing capabilities of MacinTalk plus providing several other functions to make programming easier.
SpeechLib Routines:
The SpeechLib library is the interface between your BASIC program and MacinTalk. The following BASIC commands will initiate various SpeechLib routines which will then access MacinTalk:
• CALL Speech (1[,"Exception File"] )
Loads the speech driver. The "Exception File" parameter specifies an exception file created with the exception file editor distributed by Apple in the May software supplement. If the "Exception File" parameter is omitted, MacinTalk will use its default English input pronunciation rules. If "noReader" is used as the exception file parameter, MacinTalk will be limited to phonetic code input.
• CALL Speech (2,"String to be spoken")
The default mode is English input, where the English text in "String to be spoken" is converted into a phoneme string which is then passed to MacinTalk for pronunciation. If direct Phonetic input was specified by using 'noReader' as the exception file (see above), then "String to be spoken" must be a phoneme string to be passed directly to MacinTalk.
• CALL Speech (3)
Closes the speech driver. It is important that this call be issued before ending your program, since it deallocates locked blocks of memory.
• CALL Speech (4, RATE) ;(85 ≤ RATE ≤ 425)
Controls the number of words spoken per minute. The allowable range is from 85 to 425 words/minute.
• CALL Speech (5, PITCH) ;(65 ≤ PITCH ≤ 500)
Sets the baseline pitch in hertz for the speech waveform. The baseline pitch is constrained to lie between 65 and 500 Hz.
Sets the "MODE" with which the speech will be generated. The mode can be either 0 for "Natural" speech or 1 for "Robotic" voice.
• CALL Speech (7,SEX)* ;(reserved for future use by Apple)
• CALL Speech (8,LANGUAGE)* ;(reserved for future use by Apple)
• CALL Speech (9)
Resets all the Speech parameters to their default values.
{RATE:150 PITCH:110 MODE:Natural SEX:Male}
• CALL Speech (10,"Text for Translation",Phoneme$)
This can only be used if English input is in effect. "Text for Translation" is an English string. Phoneme$ is a string variable which will contain the translated phoneme codes upon return.
• CALL Speech (11)
Provides spoken information about SpeechLib.
* These two routines are reserved by Apple for future use within MacinTalk. They will not be functional until Apple "fleshes them out" in a future version of MacinTalk.
Phonetic Input:
The MacinTalk driver produces speech from phonetic codes. MacinTalk can translate English input into phonetic codes, but errors in translation cause the nuances of speech to become lost. Optimal use of MacinTalk for fine tuning of speech is achieved by direct phonetic code input. To fully understand the direct phonetic input capabilities, the user should be familiar with the MacinTalk v1.1 manual from Apple Computer. This reproduction of the MacinTalk Phoneme Table is included for quick reference.
MACINTALK PHONEME TABLE
=============================
VOWELS
---------------------------
IY bEEt IH bIt
EH bEt AE bAt
AA hOt AH Under
AO tAlk UH lOOk
ER bIRd OH bOrder
AX About IX solId
AX and IX should never be used in stressed syllables
DIPHTHONGS
---------------------------
EY mAde AY hIde
OY bOIl AW pOWer
OW lOW UW crEW
CONSONANTS
---------------------------
R Red L yeLLow
W aWay Y Yellow
M Men N meN
NX siNG
S Sail SH ruSH
F Fed TH THin
Z haS ZH pleaSure
V Very DH THen
CH CHeck J JuDGe
/H Hole /C loCH
B But P Put
D Dog T Toy
G Guest K Camp
SPECIAL SYMBOLS
---------------------------
DX piTy (tongue flap)
Q kitt_en (glottal stop)
RX caR (postvocalic R and L)
LX caLL
QX silent vowel
UL = AXL IL = IXL (contractions, see text)
UM = AXM IM = IXM
UN = AXN IN = IXN
Stress marks
--------------------------
Nouns 5
Pronouns 2
Verbs 4
Adjectives 5
Adverbs 7
Quantifiers 7
Exclamations 9
Articles 0 (no stress)
Prepositions 0
Conjunctions 0
Secondary stress 1, 2
Punctuation
---------------------------
. sentence terminator
? sentence terminator
- phrase delimiter
, clause delimiter
( ) noun phrase delimiters
Hints for usage:
• The Speech 1 and Speech 3 calls must be paired in order for your program to function correctly. Before any speech commands can be issued, Speech 1 must be called and once speech is done, Speech 3 should be called to remove the code segments from memory.
- If speech is used consistently throughout your program ,the speech driver should be installed (Speech 1) at the beginning of the program and closed (Speech 3) only at the end. (The speech driver must find an available block of memory each time it is installed. If the heap becomes fragmented between installations of the speech driver, its possible that loading will fail.)
- If speech is used only in widely separated and selective parts of the program, then Speech 1 and Speech 3 calls should be used locally those areas. (The speech driver takes up memory and remains resident until closed.)
• Both constants and variables can be used as arguments to the speech routines. This leaves the programming options wide open:
CALL Speech 2, (TextArray(2) + Name$)
etc...
• SpeechLib is written to accept any numeric size/type (e.g. single or double precision) as the input for the speech parameters. The library will take care of converting the number to an integer for internal use.
• The use of a DEFINT statement has some side effects which must be taken into account:
If DEFINT a-z is used at the beginning of the program, you must remember to specify a larger numeric variable type for the library routine name, otherwise the program will crash. (BASIC stores the address of the machine language routine in a variable with the same name as the library routine. An integer variable can't hold a complete 32 bit address so the address is invalid.)
DEFINT A-Z
Init=1
Call Speech# (Init)
• English conversion should only be used when the string to be spoken is not known in advance (e.g. keyboard input). Direct phonetic input should be used as much as possible to insure maximum listener recognition.
• In those instances when English input and translation is used, two methods are available to correct mispronunciation. A MacinTalk exception file can be created with the exception file editor from Apple or the English spelling can be adjusted to more accurately reflect the pronunciation. (e.g. "Willey" becomes "Why lee")
• To aid program readability and ease of programming, I recommend the following opening program assignments:
Init=1: Say=2: Quit=3: Rate=4: Pitch=5:
Mode=6: Defaults=9: Translate=10: Info=11:
Natural=0: Robotic=1
• Using these assignments and the implied CALL structure (see the "SpeechLib" device.CALL statement in the BASIC manual for more information), you can access speech in a very straightforward manner:
LIBRARY "SpeechLib" this opens up SpeechLib for BASIC access
A$="Translate This" a variable assignment for a string
Speech Init,"NewRulz" initialize speech with the exception file "NewRulz"
Speech Info pronounces a 'canned' informational announcement
Speech Rate,100 sets the speaking rate to 100 words/minute
Speech Pitch,245 sets the speaking pitch to 245 hz.
Speech Mode,Robotic sets the speaking mode to Robotic
Speech Say,"Hello" pronounces the word 'Hello'
Speech Defaults resets all the speech parameters to their default
values (RATE:150 PITCH:110 MODE:Natural)
Speech Say,A$ says the English string stored in A$
Speech Translate A$,B$ takes the English string in A$ and returns the
Phonetic string in B$
Speech Quit closes the speech driver
LIBRARY CLOSE closes the link between BASIC and all libraries
If you have any questions, problems or comments, please contact me:
Kevin J. Willey home phone: (513) 426-2329
4453 Stonecastle Dr. #1102
Dayton, OH 45440
DELPHI : KWILLEY
GEnie : K.WILLEY
This project would never have been completed were it not for the support and direct help of Jim Hopper and the entire Apple-Dayton User's Group in Dayton, Ohio. If you plan on using SpeechLib in a commercial venture please contact me first. If any of you are feeling generous, you can send me what you think this is worth, a few bucks, a blank disk, some of your source code, or just a few comments. I promise in return to make every effort to support all worthwhile suggestions which come to my attention. Future efforts will be directed at the adaptation of the SpeechLib interface into the BASIC "virtual device" format. The PRINT# command could then be redirected to the
