Ming-UniAudio
Inclusion AI's unified speech model — understanding, generation, and editing in one
Inclusion AI's unified speech model combining understanding, generation, and editing.
Voice AI: lifelike TTS, accurate ASR, and real-time voice agents.
Key features
- Unifies speech understanding, generation and editing in one framework
- Unified continuous speech tokenizer merging semantic and acoustic features
- First instruction-guided free-form speech editing without timestamps
- End-to-end speech LLM with diffusion head for quality synthesis
Good for
- Researchers studying unified speech tokenizers
- Teams editing speech via natural language instructions
Stack & tags
speech-llmspeech-editingspeech-tokenizerunified-modelinstruction-guidedopen-weights