Tutorial how to create a new voice sample

[FotieMConstant]

Hi everyone, amazing project. I love it, I was wondering if there is or will be a sort of tutorial notebook on how to create a personalized voice. sort of like fine-tuning on new sample.

thanks.

[marentwickler]

I think that it's not easy/possible to fine tune it currently because the model training source is not public

[ugotworms]

Whenever they may add the additional voice packs that are available in the v.23 hugging space (such as BM Upsilon), to the repo in .pt files, how can we convert them to an 'additional_voices.json', as you had created voices.json from those files?

[thewh1teagle]

Whenever they may add the additional voice packs that are available in the v.23 hugging space (such as BM Upsilon), to the repo in .pt files, how can we convert them to an 'additional_voices.json', as you had created voices.json from those files?

With this script
https://github.com/thewh1teagle/kokoro-onnx/blob/main/scripts/fetch_voices.py

[ugotworms]

thanks, bm_lewis was good enough for my idea https://github.com/ugotworms/professor-kokoro-radio but that bm upsilon voice would sound even better

[scalar27]

thanks, bm_lewis was good enough for my idea https://github.com/ugotworms/professor-kokoro-radio but that bm upsilon voice would sound even better

Just checked it out. Very cool!

[scalar27]

Whenever they may add the additional voice packs that are available in the v.23 hugging space (such as BM Upsilon), to the repo in .pt files, how can we convert them to an 'additional_voices.json', as you had created voices.json from those files?

With this script https://github.com/thewh1teagle/kokoro-onnx/blob/main/scripts/fetch_voices.py

This makes a voices-v1.0.bin file for me. Is it still possible to get a voices.json instead?