Who did't know about Parole the speech engine for the Philips Musicmodule?
Is this program public domain? Is the developer of this program still active in the MSX scene. How did he program the fonetic data? Did he cut samples into little fonetic pieces, (I think he did, because of the french accent)
If this software PD, then I like to chance the voice into the voice of a woman.
Who can give me advice?

It's a really old program. It was distributed on many BBS's in the early 90's.. It probably was mentioned in magazines like MSX Club Magazine too...

I think you can consider it PD. Especially if there is no contact information in the .BAS files or manual.

yeah.. God is a DJ

Well, Wolf_ , I have correct my post, I understand your hint.
b.t.w I have made a nice drawing of a crying Wolf_ , I had the plan to place it here on MRC, but I have no url to host it, I will send it to you by email tomorrow, then you can create a nice cry, and put it on MRC, if you want.

*wonders about the pic*...

By the way, I recall a French speech synthesizer program (wonderful to try to overcome the accent )... Would that be the one you are referring to? No idea where I have it, though. I think it was downloaded from a BBS, or uh, on a disk magazine *grin* not really helping here, am I?? Hehe...


~Grauw

Quote:

How did he program the fonetic data?

I have some experience about this as I wrote a speech synthesis for MSX tR and made it sing "Mum's Little Piggies" in Finnish (Porsaita äidin oomme kaikki)

To record fonetic data, I made a simple BASIC program that first gave me a voice with PSG (so that I could sing all fonetics at same pitch) and then recorded 256 byte sample after I press a letter, that I want to record. I used small letters to record and big letters to replay. I must say that some fonetics were extremely hard to get sound right! Here are few tricks, that I used:

T: Knock MSX.
R: Move pencil over PC powersuply grill
S: Blow to microphone

These tricks ofcource vary regarding, what language fonetics you try to record. Finnish is very easy as you write it same way as you pronounce it. In some languages it might be good idea to record also combinations like "ae", "io" or something else. You may also want to add rules, how written language is converted to fonetics and where to place ascending and decending tone. I skipped this part and made it so, that you had to define a note for each fonetic separately. One string for fonetic data and one string for note data.

On replay you just need to play these small samples after each other. If you change replay speed, you can get voice up or down. To affect to speed, you can play the samples in loop or leave part of the sample not played. I also made a table to find good loop positions to get rid of unwanted clicks and to get rid of repeating consonants problem.

The result was not bad, but it sounded very robotic. It might have given more natural result, if I would have used longer samples.

~NYYRIKKI

Quote:

By the way, I recall a French speech synthesizer program (wonderful to try to overcome the accent )... Would that be the one you are referring to? No idea where I have it, though. I think it was downloaded from a BBS, or uh, on a disk magazine *grin* not really helping here, am I?? Hehe...

Duh
It's probably one and the same. Legacy already talked about the french accent, and I said Parole was distributed mainly through BBS"s

I think there may be one other, which is not as good... And ofcourse, there's the built-in SFG-module talky-basic. CALL SAY etc... Japanese accent ofcourse

Quote:

I must say that some fonetics were extremely hard to get sound right! Here are few tricks, that I used: T: Knock MSX. R: Move pencil over PC powersuply grill S: Blow to microphone

Nice
I created a snare-drum once by stacking a few boxes of diskettes on top of eachother and then throwing them over