By the by, don't know how many of you fellas are interested in this but here goes: A message from the cuckchan threads: Specifically we want to fork this text to speech synthesis project to compete with 11labs speech synthesizer. Source code for tortoise tts https://github.com/neonbjb/tortoise-tts here is the specific code we need to break down and understand https://github.com/neonbjb/tortoise-tts/blob/main/tortoise/utils/diffusion.py training models used https://github.com/neonbjb/tortoise-tts/tree/main/tortoise/models Source for the trainer project tortoise used https://github.com/neonbjb/DL-Art-School Tortoise tts videos https://www.yewtu.be/watch?v=J3-jfS29RF4 https://www.yewtu.be/watch?v=Kfr_FZof_hs https://www.yewtu.be/watch?v=FN3yxL0Rr0c we have datasets https://github.com/LAION-AI/audio-dataset Given the potentially large datasets involved it is a prudent strategy to decentralize training going forward. If you know anything about federated training or understand about these github projects can you explain which are the best and most relevant. https://github.com/topics/federated-learning https://en.wikipedia.org/wiki/Federated_learning This is new tech for many anons we need some experienced ML anons to help push this in a useful direction. Impart your wisdom anons. Other datasets and info: https://rentry.org/AIVoiceStuff