Possibly add talking Avatar?

[ling-k]

It would be more engaging if we could have a talking avatar. There are some open-source models and non-free ones as well.

[natolambert]

@ling-k do you have any examples?
Part of why I'm hesitant is that I don't see them at the same level of maturity, but happy to look at them.

[ling-k]

@natolambert I have tried the Azure AI one. It works well but not free.
https://speech.microsoft.com/portal/talkingavatar
Now, there are several avatars we can select. I think they will allow users to upload their own avatar and then retrain the generation model.

[ling-k]

Something like the video below is also pretty good. Some different avatars. Not sure what they have used.

https://www.youtube.com/shorts/RzMMPUY09SQ

[natolambert]

Unsurprisingly, most of the APIs look pretty expensive for my amount of content (~30-60min/month), and many don't allow me to bring my own audio file yet. I'm more invested in having it sound pretty close to me for now (which it is doing surprisingly well at) than the video now.

I agree it's kneecapped on YouTube essentially being a audio only file.

(I'm not opposed to adding a demo to the repo, I'd kind of want it to be pretty clear which should be supported before doing it. I suspect many companies / models will come an go in this space. 11labs seems pretty well established, but also that could die and yield annoyance)