Community
Kokoro TTS paired with OpenVoice converter gains voice cloning
A community recipe pairs Kokoro with OpenVoice's ToneColorConverter for voice cloning, adding about 600 MB peak VRAM and roughly 10% slowdown on CUDA. Speaker similarity scores of 0.7 to 0.8 were measured with CAM++ using synthetic reference voices.
Every outlet that ran it 1 article
r/LocalLLaMACommunityLead
Kokoro can now clone any voice with just ~600 MB of extra VRAM!