I need a fast direct audio to audio - Voice Conversion Model which I can finetune with my custom data in german. I found the voice_conversion.py script where piper seems to be used for direct voice conversion.
When I try using the German checkpoint it caused an Error. Do I need to do any kind of preprocessing/preparing of the model before using it in the script? Or do you have specific pretrained checkpoints for voice conversion? Unfortunately I couldn't find any hints/tutorials or readmes in using Piper for voice conversion. Also when training Piper for voice conversion with a new voice, does it differ from training for tts?
Hi,
I need a fast direct audio to audio - Voice Conversion Model which I can finetune with my custom data in german. I found the voice_conversion.py script where piper seems to be used for direct voice conversion.
Now my question: Can I use any of the Checkpoints saved here: https://huggingface.co/datasets/rhasspy/piper-checkpoints/tree/main
When I try using the German checkpoint it caused an Error. Do I need to do any kind of preprocessing/preparing of the model before using it in the script? Or do you have specific pretrained checkpoints for voice conversion? Unfortunately I couldn't find any hints/tutorials or readmes in using Piper for voice conversion. Also when training Piper for voice conversion with a new voice, does it differ from training for tts?
Thank you for your help!