Re: [Audiogames-reflector] Making A Tts Voice?
Re: Making A Tts Voice? Lol I tried model talker, and I cant mannage recording a simple phrase, because of that stupid microphone calibration. When I do it right, the program some kinda crash, and sometimes it tells me the program somehow does not recognize any voice of mine. I used my studio equipment and chosen the right mic.Even if you say it sounds crappy at the end, I may try fiddling around with it and see what comes out.URL: http://forum.audiogames.net/viewtopic.php?pid=169254#p169254 ___ Audiogames-reflector mailing list Audiogames-reflector@sabahattin-gucukoglu.com http://sabahattin-gucukoglu.com/cgi-bin/mailman/listinfo/audiogames-reflector
Re: [Audiogames-reflector] Making A Tts Voice?
Re: Making A Tts Voice? Ive found it easier to just cut up recordings and piece words and phrases back together. Sometimes, the results even sound natural (but usually you can tell its been edited). It helps to have a large database of words/phrases; cutting them up, and getting your sorta-kinda-TTS to identify which is best to use in a given situation is complicated, however.(One could presumably turn such a program into a sapi5-compatible voice with some c++ header magic and getting it to implement all the required functions, but thats more than Im willing to do for the uses Ive needed. Conceivably, you could get a programmer to handle the program itself, but it will inevitably be too dubious quality for how much anyone would charge for the headache of coding it to be sapi compatible.)URL: http://forum.audiogames.net/viewtopic.php?pid=168868#p168868 ___ Audiogames-reflector mailing list Audiogames-reflector@sabahattin-gucukoglu.com http://sabahattin-gucukoglu.com/cgi-bin/mailman/listinfo/audiogames-reflector
Re: [Audiogames-reflector] Making A Tts Voice?
Re: Making A Tts Voice? Guitarman, to be honest the requirements to do this are far more complicated and more technical than you realize. It is definitely far more complicated than just writing a scriptSupposedly there are programs like Model Talker which Scot mentioned above where you could record sample clips of yourself and construct a Sapi 5 voice, but as Scot mentioned the end results were pretty low quality. There is also an open source TTS Engine, Festival, which has a number of sample voices, but to be honest no one has come up with a decent enough voice for it which sounds even as good as the Vocalizer voices, and even if you did start with something like that it would require months of research into the APIs as well as the science behind voice synthesis. Do you really want to do that?URL: http://forum.audiogames.net/viewtopic.php?pid=168866#p168866 ___ Audiogames-reflector mailing list Audiogames-reflector@sabahattin-gucukoglu.com http://sabahattin-gucukoglu.com/cgi-bin/mailman/listinfo/audiogames-reflector
Re: [Audiogames-reflector] Making A Tts Voice?
Re: Making A Tts Voice? There was a program called Model Talker that essentially did this. They had his list of phrases, youd read them off, and youd have a SAPI-based voice. It worked, but it was very low quality. Indeed, anything you could create, frankly, would suck. It might be better than eSpeak, but again, itd be SAPI5-based, and theres plenty of insanely good ones out there already.URL: http://forum.audiogames.net/viewtopic.php?pid=168862#p168862 ___ Audiogames-reflector mailing list Audiogames-reflector@sabahattin-gucukoglu.com http://sabahattin-gucukoglu.com/cgi-bin/mailman/listinfo/audiogames-reflector
Re: [Audiogames-reflector] Making A Tts Voice?
Re: Making A Tts Voice? CAE, there are a lot of problems with that method. For one if you cut up recordings and piece words and sentences back together there is always going to be certain words or phrases that your TTS will never be able to speak. If it isnt in your database it wont be able to read it. I often see things like this happen with the ATT Natural voices, and it flat out does not work for strange words from a science fiction novel or fantasy novel where the words are often made up and need to be sounded out. Certainly they dont appear in any dictionary and a TTS voice as you are proposing just would not be able to do it.Another problem is performance. TTS engines like Espeak may not sound human but are very responsive when used with a screen reader etc. Thats because they dont have to do a lot of complex work when generating the speech. What you are proposing requires much more complex work from the CPU and memory stringing sounds together to reconstru ct a human voice. I am not at all certain you could get the same amount of performance from such a system because you are in a sense loading and playing sound clips back rather than synthesizing the speech.URL: http://forum.audiogames.net/viewtopic.php?pid=168905#p168905 ___ Audiogames-reflector mailing list Audiogames-reflector@sabahattin-gucukoglu.com http://sabahattin-gucukoglu.com/cgi-bin/mailman/listinfo/audiogames-reflector
Re: [Audiogames-reflector] Making A Tts Voice?
Re: Making A Tts Voice? Yes (this is one of the reasons I wouldnt bother with trying to make such a thing sapi compatible). I certainly wouldnt recommend it for a screen reader voice.(On the other hand, there are languages for which it would work much better, such as Japanese, where you can get everything you need by recording the alphabet in a few different tones. The overhead issues still apply, of course.)URL: http://forum.audiogames.net/viewtopic.php?pid=168998#p168998 ___ Audiogames-reflector mailing list Audiogames-reflector@sabahattin-gucukoglu.com http://sabahattin-gucukoglu.com/cgi-bin/mailman/listinfo/audiogames-reflector
