This is fun. The audio isn’t particularly believable. Not that it matters for this project.
I’ve used a few tts providers and although a technical marvel they are all still easily identifiable as ai. Humans are very good at spotting fakes.
This project might work better with mostly canned audio clips of real humans.
I disagree. I think many of the audio elements are highly believable for lay people.
Hell no it's not. It's extremely uncanny.
What would "lay people" even be in this scenario? I'd wager most people know what a meeting sounds like.
It sounds like voice acting. And not particularly good voice acting. Too clear, a bit too bombastic, and as if reading from a script. At least the two that I briefly listened to.
My grandma who has never used ChatGPT would have probably been fooled. For the rest of us though, this has got that annoying ChatGPT voice inflection that is hard to miss.