plugin for TTS (text to speech) integration

Hussayn wrote on 5/8/2008, 4:40 AM

Hi;
I am creating video tutorials with vegas-8. Recently i started using a text to speech system for creating the voices. That went very well (see http://machinimatrix.blip.tv for 2 examples) , but was also very time consuming, because i had to create the entire voice over in one shot, so i needed to tweek timing by hand.

My idea is to write a script(?) which allows me to "insert generated TTS voice" into an audio track. I would edit the text right within the plugin. The plugin knows how and when to call the TTS system and generates the voice as needed (and of course holds it in the cache, so regeneration will be done only after a text change is detected).

My questions:

1.) Maybe such a plugin allready exists ? Where would i have to search for that ?
2.) Is it at all possible to write such a plugin and integrate it seamlessly into Vegas, like for instance Cinescore has been integrated ?
3.) Where would i get information about how to start such a scripting task (if it is possible at all) ...
4.) Is there any common interest in such a tool ?
5.) If 4.) applies, where would i eventually have to offer my finished script ?

thank you for any hint / tip / suggestion.

regards, hussayn

Comments

JohnnyRoy wrote on 5/11/2008, 4:25 AM

I am not sure about the kind of integration you are looking for. If the voice generator created audio files for each sentence, then you could write a script that searches for markers and places an audio file at each marker. To use it, you would place markers where you want the speech and the script would add the audio at that position.

As far as the kind of integration like Cinescore, your script would need the ability to call the text to speech system and then place it's output at the current cursor position. I'm not sure if the text to speech system can be called like this but if it can, this might be possible. I could help you figure it out.

~jr

Hussayn wrote on 5/11/2008, 11:26 PM

hi.
thank you for your response.
We have created a Web service around our TTS system, which replies with an audio file, when we send written text to it, independent of its length. So the plugin, we are thinking about is not at all specialised on our TTS, but on a "content generator accessible through an URL". It could basically be written in a way, that the generated content could come from any URL in the world. It even could be a video, although currently i don't see, where that make sense ;-)

So in principle, what we need to find out is, how can we create a script, that integrates like cinescore. We would enter text into a text box, the script would then call the TTS-URL, receive the audio and place it into the audio-track. Later we can move the generated object along the time line with the mouse.

From your post i conclude, such a plugin can be implemented?
So, how would we start with that script ?

I am an experienced software developer (mainly java and c/c++), but totally new to vegas scripting. So at the moment i am a bit blind folded, since i could not find any good tutorial yet. The language reference does not help much to get the basic principles. So if you could give me some starting hints, that would be great.

A plugin implemented on the base of "c# sounds a good approach ?

jetdv wrote on 5/12/2008, 5:46 AM

Here's one thought that might work. You could put a series of regions or markers on the track and name them the text you want said. Then a script could easily read those names, send that text to your TTS system, it would have to know how to get the resultant file, but then it could put that new file on the timeline automatically.

There's no reason why you couldn't type text directly into the script either. Was just thinking putting them all in ahead of time might be a time-saving factor.

The catch is properly sending the text to your TTS system and knowing how to get the resultant file.