Spaces:
Running
TTS EVALUATION REQUESTS
Request open-source TTS models for evaluation here.
Most interesting ones seem to be in to be evaluated already but this recent one would be nice too
https://huggingface.co/ResembleAI/chatterbox
Thank you for the requests
@Gatozu35
!
We are including chatterbox in the evaluation - we unfortunately cannot include Kokoro since it isn't voice cloning.
We don't expect listeners to do well in ranking TTS systems that do not use the same speakers, so we don't think our objective evaluation would do well either - there are just too many unknown variables in that case.
Thank you for your response. I understand your decision. The main reason I wanted kokoro to be evaluated is that it usually scores very well in metrics even though the prosody is subpar subjectively. I guess I'll try computing the prosody metrics locally.