Can it replace Elevenlabs?

rutvik106 commented 8 months ago

I just want to know that is it possible to use this tool on my AWS Server to create voices for my voice cloning product? Currently I am using Elevenlabs to do it. And it is very costly.

H46K3RL0L commented 8 months ago

Don't quote me on this, but it should be possible with some changes to the code. I think these would be pretty big changes though because this is voice CONVERSION, not voice GENERATION.

78Alpha commented 8 months ago

It needs to be combined with a TTS tool. This just applies the voice properties to audio. You have some options for TTS with ups and downs.

Tortoise (Old, slow, but allows for synthesizing completely new voices and cloning existing voices with accent)
XTTS (Clone a single voice or multiple voices as different speakers, including accent)
Piper (Clones a voice, no emotion and slow to train)
StyleTTS (Can learn the style of a voice and clone it, allowing for control of emotional inflection, but needs 24 GB VRAM+ for training)
SUNO (It can make voices, lots of features, no real voice cloning. A voice can be cloned but not trained, so it will have poor quality)
Built-ins (AWS, Azure?, Edge, so on)

github-actions[bot] commented 4 months ago

This issue was closed because it has been inactive for 15 days since being marked as stale.

RVC-Project / Retrieval-based-Voice-Conversion-WebUI

Can it replace Elevenlabs? #1667