Kokoro gets voice cloning with ~600 MB extra VRAM via OpenVoice's ToneColorConverter
Acceptable-Cycle4645 · reddit · 2026-10-30
Inspired by OpenVoice's TTS + voice conversion design, a developer paired OpenVoice's ToneColorConverter with Kokoro in a new 'any-tts' recipe for audio.cpp (recipes introduced in v0.9.2), adding voice cloning to existing TTS models without retraining. Overhead: 600 MB extra peak VRAM, 10% slowdown on CUDA, 0.5x speed on CPU but still 2x real-time. The TTS handles content, pronunciation, and prosody while the VC model transfers the target speaker's timbre — applicable to any tiny TTS model.
More from Multimodal
- HyperKut Studio: open-source multi-shot AI video studio on ComfyUI — Main_Creme9190 · 2026-10-30
- Engineers with Claude act as the marketing department for Sensmore print ads — MarvinTBaumann · 2026-10-30
- User says Higgsfield wiped her paid subscription and denied it; lists alternatives — aziz4ai · 2026-10-30
- Pika Labs open-sources H3 Flex, a MiniMax H3 finetune with 6-step generation and 12 reference inputs — LatentSpacer · 2026-10-30
- Comfy Cinematic: open-source mobile browser client for ComfyUI with auto workflows — No-Sleep56 · 2026-10-30
- Viral bobblehead-hugging-giant-Snickers image prompt delivers cute results in seconds — aitrendz_xyz · 2026-10-30