As of 28 July, this podcast has dramatically declined in quality. Instead of human narration, it is now “Narrated for LessWrong by TYPE III AUDIO”. It’s a text-to-speech model that has far too many problems for this to be listenable. The prior human narration was excellent, e.g. the effort put into describing images.
Please revert this change, or move it to a different podcast!
I do definitely agree human vs machine should be separate podcasts.
The decision to have the auto-reading is pretty different from the decision to have individual posts get custom human narration (we’re a bit more funding crunched now and it was less obviously worth it for every individual curated post. but the automated stuff is cheaper and easier to just do for a large swath of posts). But we’ll keep feedback like this in mind when deciding how much to prioritize each of them.
As of 28 July, this podcast has dramatically declined in quality. Instead of human narration, it is now “Narrated for LessWrong by TYPE III AUDIO”. It’s a text-to-speech model that has far too many problems for this to be listenable. The prior human narration was excellent, e.g. the effort put into describing images.
Please revert this change, or move it to a different podcast!
Thanks for the feedback.
I do definitely agree human vs machine should be separate podcasts.
The decision to have the auto-reading is pretty different from the decision to have individual posts get custom human narration (we’re a bit more funding crunched now and it was less obviously worth it for every individual curated post. but the automated stuff is cheaper and easier to just do for a large swath of posts). But we’ll keep feedback like this in mind when deciding how much to prioritize each of them.