koelab is a local voice conversion and TTS workbench — “Ollama for Voice” — unifying voice conversion, Japanese TTS, multilingual TTS, WebUI workflows, and real-time CLI workflows in one project.
Key Features:
- Voice Conversion — convert source audio using a target reference voice
- Japanese & Multilingual TTS — synthesize speech from text with multiple model backends
- Real-Time CLI Workflows — live voice conversion from microphone input
- WebUI + REST API — browser-based interface and programmatic access
- Model Registry — pull, list, search, and inspect models with simple commands
- Local-First — practical local alternative to hosted services for control, privacy, and hackability
- Docker Support — containerized deployment with
docker compose
Quick Start:
pip install koelab
# Pull a voice model
koelab pull seed-vc
# Convert voice
koelab run seed-vc --input source.wav --reference target_voice.wav --output converted.wav
Core Commands:
| Command | Description |
|---|---|
koelab pull <model> |
Download a voice model |
koelab run <model> |
Run voice conversion |
koelab speak <model> |
Synthesize speech from text |
koelab live <model> |
Real-time voice conversion from mic |
koelab serve |
Start REST API server + WebUI |
koelab doctor |
Inspect runtime dependencies and setup problems |
koelab models |
Show all available models from registry |
Technologies:
- Python 3.11+ (3.12 recommended)
- Voice conversion models (seed-vc, etc.)
- TTS backends (Irodori-TTS, XTTS-v2)
- FFmpeg integration (recommended via autoffmpeg)
- Docker / docker compose
Use Cases:
For developers and creators who want a local, hackable alternative to hosted voice services. Ideal for voice conversion experiments, multilingual TTS workflows, real-time voice transformation, and privacy-conscious audio processing.