Well anons, I guess it happened!
It only took less than a week for ElevenLabs to do something retarded and kill their userbase. Glad you guys had fun, but it’s time to start true dataset creation work for open-source alternatives.
As I mentioned a few threads ago, I hope you guys have been keeping your source audio in high quality and in high amounts. Start looking into setting up systems for collaborating on further collection, transcription and curation of audio. /mlp/ and the Pony Preservation Project has a wealth of documentation on what format everything should be in, plus many guides to ease the dataset creation process. We have well over 5 years worth of nigh-constant streaming to go through, so get in contact with your oshi's general boards and get as many hands on deck as possible. It's not going to be easy but we know it's possible.
I’ve already submitted small datasets for Anya, Ollie, Kronii, IRyS, and Sana to
15.ai months ago (which I’ll probably share the source audio for soon), but don’t feel like you don’t have something to contribute, because datasets can always be improved. Archive. Transcribe. Curate. If a bunch of horsefuckers could spearhead the voice AI revolution three years ago, there's nothing stopping one of the fastest boards on 4chan from doing so now.
We’ve all had an ample taste of what voice AI is capable of. The time is now.
>inb4 nypa 