Better practices
Hi,
Very thank you for this Space!
I am wondering: How much time of recording is optimal? Is there a way to clone a voice in other languages?
Yes, but first we need to gather data(recorded voice chunks with their labels) and train a neural network on the target language.
Great! And how much time of recording voice is optimal for cloning in this model?
Thank you!
To be honest, I have no idea. I have tried it with 30 to 40 seconds and it worked well. Its a quick cloning model so users are not supposed to record much for it and the cloned voice will of course not on point. For better results we can try neural networks which take upto an hour long recording.
Dear Bilal,
do you have any idea where we can find such a neural network, open source?
https://github.com/coqui-ai/TTS/blob/dev/notebooks/Tutorial_2_train_your_first_TTS_model.ipynb
https://github.com/jaywalnut310/glow-tts/issues/33
@fspecii
check this. You might find it beneficial.