Speech Synthesis

[UseFull Apps]

how do you do speech synthesis with xna like the app Talking Clock,
i would just look at the source code but i lost that file and i can not download it again.. so any help would be welcomed

[Tiptup300]

Originally Posted by UseFull Apps

how do you do speech synthesis with xna like the app Talking Clock,
i would just look at the source code but i lost that file and i can not download it again.. so any help would be welcomed

You can't really do real speech synthesis, Talking Clock just used a saved list of samples for various things like.

"One" "Two" "O'clock' "Thirty"

[UseFull Apps]

kk i thought so cause i could not find anything on speech on google

[Shadow]

Originally Posted by Tiptup300

You can't really do real speech synthesis, Talking Clock just used a saved list of samples for various things like.

"One" "Two" "O'clock' "Thirty"

Couldn't you (not quite simply because it could be difficult, but there's no other word for it) simply record all of the sounds of the human voice and make the program play them back in order accordingly?

[Tiptup300]

Originally Posted by Phantom707

Couldn't you (not quite simply because it could be difficult, but there's no other word for it) simply record all of the sounds of the human voice and make the program play them back in order accordingly?

Sure you have all the space in the world. Sarcasm aside though it sounds terrible. You don't say words the same exact way in sentences. "Live" is a good example.

[BroiledVictory]

Its possible. Is it worth it? No. It would be very hard.

[Tiptup300]

Originally Posted by SmileDog

Its possible. Is it worth it? No. It would be very hard.

I hate your signature so much. Please burn it.

[God]

Originally Posted by Phantom707

Couldn't you (not quite simply because it could be difficult, but there's no other word for it) simply record all of the sounds of the human voice and make the program play them back in order accordingly?

That's basically what Voice Synthesis programs do, break words into sounds, play sounds in the correct order, with basic grammatical gaps. Hence why they always lack tone

[Tiptup300]

Originally Posted by Nurta

That's basically what Voice Synthesis programs do, break words into sounds, play sounds in the correct order, with basic grammatical gaps. Hence why they always lack tone

Certain ones, not many of the new ones.

[God]

Originally Posted by Tiptup300

Certain ones, not many of the new ones.

Well yea, now with faster processors, have it completely computer generated isn't too hard, but the basic ones, especially if they have a high storage/processor ratio, like the Zune.

But yeah to reiterate the past posts, if you want to do basic speech synthesis, you could on the Zune, but best to do it the simplest way, and do metaphorical equivalent of raycasting, it'll save yourself some headache. Break it up into as few blocks as possible like the clock and play them in order.

[Tiptup300]

Originally Posted by Nurta

Well yea, now with faster processors, have it completely computer generated isn't too hard,

There still aren't any really good voice synthesizers. I doubt it's processing power. It's just that the computer has to be able to understand a sentence to read it and I doubt that's easy to do.

[badmoodguy88]

There still aren't any really good voice synthesizers. I doubt it's processing power. It's just that the computer has to be able to understand a sentence to read it and I doubt that's easy to do.

They are a lot better than you might think Cepstral is good they sell by the voice and so it is fairly cheap. Callie I think sounds vary understandable and somewhat expressive. It does a thing were it puts emphasis on certain words that almost always have an emotional charge to them so that the text is read with some feeling.

ATT also has a voice synth that you can try online that is very good and very expensive. Some of the voices sound very real but grammatically (putting emphasis in the right places) it is not as good as Cepstral.

You could record the phonic sounds but then the challenge becomes the logic that recognizes the exceptions like "live" and program specifically for them OR if you do have it for a game you could have a text string that gets displayed "live" and another that gets read by the program 'liv'

The doable method would be to restrict your self to a limited vocabulary and just reuse words saved.

The zune games can be HUGE, the thing is that you can not load that much of it at a time...

[Cobalthex]

if this has not already been said (didnt read comments above )
you can in theory do speech synthesis, its just extremely tedious and probably not very easy on XNA (not that it is period) and it not worth the reward