Youssef Rizk: belief

Transcript context

…Stories and like, what are you looking for? What is like the holy grail? What, what is, what are you looking for? Just honestly how human it sounds and like how likely I'd be to listen to this if I, if I did it. Also, how like customizable it is. Yeah. I think the problem with all these voice things and generally a lot of the as stuff is it is somewhat random. Mm-hmm. But you're using it in production applications that require certainty, right? Just as an example, if I promise my users this podcast or this segment will be 30 seconds, it needs to be 30 seconds. Or, you know, given some SLA. Discontolerance. Some SLA around, like, you know, it's 95 percent of that. But I think a lot of these things just tend to be a little random at the moment. So like how, can I, can I literally specify a tone that I'd like this to read it in and be certain that it's doing it and it's not some weird like attempt at sounding surprised? Yeah. It's just like, yeah, basically how controllable and how realistic they sound. Yeah. And then final, final question around just the landscape of TTS. What are the unique challenges for non English TTS? And I'll tell you, right? So, I'm interested in having 28 languages of latent space, right? That's only good things for me. Except if it sucks. And I obviously have no way to validate. I think that's the problem with Latent space Ukraine. Yeah,…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.