I want to make a speech synthesis program. Of the kind that reads text to you. I think I have an idea of how to do this. I just need a couple bits of information (maybe megabits ) Where can I get a list of every sound in the English language and the rules of when to use it (aka thing like "I before E except after C unless sounding like AY as in neighbor and weigh"). You know... all those rules that the English teacher makes you learn in elementary school. And I need the sounds so I know what to record as the WAVs. Second question: Is the PlaySound() function (I think thats the name) fast enough so that when you play a sound in blocking mode (the opposite of asynch?) and immediately play another one, can the user hear a pause while it loads the sound from a file? I want the computer to say "Hello" not "H-e-l-o" --------------------

You are not a real programmer until you end all your sentences with semicolons;

Yanroy@usa.com

Visit the ROAD Programming Website for more programming help.

You want phenomes :

PhonemeList = ( 'U', 'A', ' ', 'B', 'D', 'G',
'J', 'P', 'T', 'K', 'W', 'Y',
'R', 'L', 'M', 'N', 'S', 'V',
'F', 'H', 'Z', 'AW', 'AH', 'UH',
'AE', 'OH', 'EH', 'OO', 'IH', 'EE',
'WH', 'SH', 'TZ', 'TH', 'ZH' );

Instead of rules, programs like this usually use dictionaries of how the phenomes are strung together ... Look here for one :

ftp://ftp.cs.cmu.edu:project/fgdata/dict
(Latest seems to be cmudict.0.3.Z)

And don't use wavs - you can generate the phenomes on the fly ... even through the PC speaker

Edited by - morfe on September 7, 2000 9:46:18 AM

"I want to make a speech synthesis program. Of the kind that reads text to you."

Hehe, remebers me of the good ´ol dos-soundblaster 2.0-talker....How funny it was to let it read German Texts
(It was the English Version of the app - you understand).

Greets,XBTC!

i don''t think PlaySound() will always break up the sounds, but it seems likely that it''ll happen a bit.. think about it: it loads a sound, plays it, releases it, loads, plays, releases..
to make the h-e-l-l-o thing not an issue, you could possibly compose it all to some kind of buffer before you play it.. that''d work great for short phrases/words, probably not so great for paragraphs..

-------------------
LPUNKNOWN pUnkOuter

You want to build a data stream and send it to the sound card continuously. It''s a great excercise anyway (finally wrote a library class that does it for me). Look at using the waveOut family of win32 functions (waveOutOpen, waveOutPrepareHeader, waveOutWrite, waveOutClose). Fun stuff.

quote: Original post by morfe

Look here for one :

ftp://ftp.cs.cmu.edu:project/fgdata/dict
(Latest seems to be cmudict.0.3.Z)

There doesn''t seem to be anything at that address. I tried just ftp://ftp.cs.cmu.edu but the folder project was empty. Actually, all the folders were empty.

quote:
And don''t use wavs - you can generate the phenomes on the fly ... even through the PC speaker

How would I go about doing this?


--------------------


You are not a real programmer until you end all your sentences with semicolons;

Yanroy@usa.com

Visit the ROAD Programming Website for more programming help.

I know it''s a really evil thing to do, but I really need an answer, at least to where I can get that phenome list So I am just posting this to keep the thread alive.

--------------------


You are not a real programmer until you end all your sentences with semicolons;

Yanroy@usa.com

Visit the ROAD Programming Website for more programming help.

Ok, you evil person you, look here first :

http://www.cgocable.net/~jrussel/voicelab.htm

-------------------------------------------------
Mindphuq Software : "Who do you want to do today?"

And while you're at it, you might as well have a look here, mainly for its implementation of ASCII phenomes with RSynth (which is free to D/L) :

http://wwwtios.cs.utwente.nl/say/

Edited by - morfe on September 9, 2000 11:27:13 AM