Very nice to see this as a package. Maybe I'll give it a gander.
I have this personal, related story. One day I was walking my dog, and I'm trying to learn Spanish, alright? So there I was, walking my dog, listening to Game of Thrones audiobooks, all in Spanish. Not because I'm some hardcore fantasy fan, I just figured since I already read the books and watched the show, I'd catch enough to make it some good language practice.
But then I started missing chunks of the story (I'm still at the very beginning of my language learning journey) and got curious about what was actually happening. So naturally, I turned to ChatGPT: "Hey, can you tell me what happens in this exact scene, can you read that passage from the book?", in English, of course.
ChatGPT hits me with the classic: "Sorry, copyrighted material, can't quote it verbatim." Fine. "Give me a Google query to find this passage then."
That led me down a rabbit hole trying to find the exact text, but I got impatient and was about to give up - I had the books somewhere, I'll try to find the exact chapter later, I thought. However, I stumbled onto some random blog where this person was analyzing GoT from a historical perspective. Actually pretty fascinating stuff - comparing fictional events to real historical battles, etc. I got hooked on it for a while.
But then it hit me. Wait a minute. Five minutes ago I was listening to some shit, now I'm forced to read some other shit? Why am I going backwards? "No, no, no! You! Yes, you, mister computer, please read this shit to me!"
So I tried the Mac's built-in 'say' command. Meh. It sounded like a robot having an existential crisis – so fucking unbearable.
I thought, "Oh, come on! We have these amazing AI models today that can sound almost human." So I went hunting through TTS engines – Mozilla thing, some other options – and then I found Piper. Oh shiiit, this thing actually sounds really good!
For some reason I can't quite remember, I couldn't get it running natively on Mac, so I stuffed it into a Docker container. Next thing I know, I'm writing Emacs Lisp to make it all work seamlessly in my editor.
And you know what? It's actually really nice. I loved it. Even though I haven't used it much since then. Sometimes the most interesting experiments come from the weirdest chains of procrastination.
Very nice to see this as a package. Maybe I'll give it a gander.
I have this personal, related story. One day I was walking my dog, and I'm trying to learn Spanish, alright? So there I was, walking my dog, listening to Game of Thrones audiobooks, all in Spanish. Not because I'm some hardcore fantasy fan, I just figured since I already read the books and watched the show, I'd catch enough to make it some good language practice.
But then I started missing chunks of the story (I'm still at the very beginning of my language learning journey) and got curious about what was actually happening. So naturally, I turned to ChatGPT: "Hey, can you tell me what happens in this exact scene, can you read that passage from the book?", in English, of course.
ChatGPT hits me with the classic: "Sorry, copyrighted material, can't quote it verbatim." Fine. "Give me a Google query to find this passage then."
That led me down a rabbit hole trying to find the exact text, but I got impatient and was about to give up - I had the books somewhere, I'll try to find the exact chapter later, I thought. However, I stumbled onto some random blog where this person was analyzing GoT from a historical perspective. Actually pretty fascinating stuff - comparing fictional events to real historical battles, etc. I got hooked on it for a while.
But then it hit me. Wait a minute. Five minutes ago I was listening to some shit, now I'm forced to read some other shit? Why am I going backwards? "No, no, no! You! Yes, you, mister computer, please read this shit to me!"
So I tried the Mac's built-in 'say' command. Meh. It sounded like a robot having an existential crisis – so fucking unbearable.
I thought, "Oh, come on! We have these amazing AI models today that can sound almost human." So I went hunting through TTS engines – Mozilla thing, some other options – and then I found Piper. Oh shiiit, this thing actually sounds really good!
For some reason I can't quite remember, I couldn't get it running natively on Mac, so I stuffed it into a Docker container. Next thing I know, I'm writing Emacs Lisp to make it all work seamlessly in my editor.
And you know what? It's actually really nice. I loved it. Even though I haven't used it much since then. Sometimes the most interesting experiments come from the weirdest chains of procrastination.
——
https://github.com/agzam/dockerized-piper