BREAKING
Qwen3-TTS Arrives in ComfyUI
0
Languages
0
Preset voices
0
s
Max reference
Two Model Sizes
1.7B
High quality
●
Best fidelity
●
More VRAM
0.6B
Lightweight
●
Lower resource use
●
Faster
How Voice Cloning Works
1
Reference clip
↓
2
Whisper transcribe
↓
3
Clone voice
↓
4
Generate speech
Watch the Setup Limits
Cloning Meets Local Workflows
AI NEWS BLITZ
Alibaba's open-source Qwen3-TTS now runs inside ComfyUI with voice cloning.