Impressive, I wrote a lib to do something similar but I didn’t know how to make it in two different voices
using podcastfy (python lib) I just created a workflow where I can turn any url, pdf or youtube video into a custom ai-generated podcast. it even generates an rss feed so I can listen on my phone via fountain when I'm out and about.
I just pointed it at my relatively empty website to see what would happen:
https://cdn.jb55.com/s/bbbd6633c2fafa9c.mp3
this is hilarious.
Discussion
Not sure what podcastfy does, line by line I guess? I noticed i hit the api rate limit on openai tts-hd, not sure why though since i seem like im still under RPM limit…
Looking into kokoro now but im not sure if that supports different voices
I gave it a go and its indeed good.
The reason I wrote the code for that is I wanted to convert NIPs readme into some easy to listen / short podcast series that new coming devs could listen on their way back from the office and get a general idea on which nips to use for what purpose.
mine: https://hook.cafe/files/mylib.mp3
podcastify: https://hook.cafe/files/podcastfy.mp3
obviously they win with their voice switch, but I already uploaded mine to apple during the weekend just for a test and I didn't know that python lib existed! Anyways, maybe this will help someone one day:
https://podcasts.apple.com/us/podcast/walk-through-the-nips-of-nostr/id1799447906