TTS Developments

I've continued forward with Oculus Voice for TTS (after finding a wonderful recorded livestream of a developer showing how he set it up). The documentation through Oculus, Unity, and Wit.ai (the online server/engine used to produce the actual TTS) is very scant, and getting it running using the few words they offer about it was very difficult. Luckily, thanks to Aaron Jackson, I now understand how the different components work, and how they can be used in script to produce TTS. It even comes with a couple voices with different emotions, which is really great! (I hate that they're all American accent voices, haha, but that's beside the point.)

I've also got in contact with ReadSpeaker.ai, and they've provided me with their Unity Package. It currently is throwing errors upon installing, so I've reached out to their support to see what can be done. It also sounds like, because this is a trial package, the audio will be watermarked which is incredibly annoying and immersion breaking. So that might be my absolute last minute solution. Meh.

So, I'll show a little about how the TTS works.

I've got my (incomplete) model here. His name is Raith. He's fully rigged up, and the only remaining steps to complete with him are the textures and all the BlendShapes / ShapeKeys for his facial expressions.

The Oculus SDK comes with a TTSSpeaker component, which gets added to the character. The Speaker has an audio source and voice presets attached.

From there, the speaker gets called from a script where you can speak whatever String input is sent to it. This is a really crucial element for my project, as once I get the OpenAI chat part working, I'll need to feed its results to the TTS through script.

Finally, we get this!

From here I'll work on getting the Chat bot set up. Then all of the facial expressions for Oculus Lip Sync.

Here's another little thing that I love doing with all my 3D characters. Flowy hair. hehe. Also, a little BlendShape experiment to make sure it works.