ComfyUI Node
Qwen3-TTS Voice Clone Prompt ⚡
Extract and cache voice features from reference audio for reuse in the Voice Clone node. Uses a *-Base checkpoint, loaded in-node.
Qwen3-TTS Voice Clone Prompt ⚡
- ref_audio
- voice_clone_prompt
◄ref_text►
◄model_choice0.6B►
◄deviceauto►
◄precisionbf16►
◄attentionauto►
◄x_vector_onlyfalse►
◄unload_model_after_generatefalse►
Category🤖 CCTech/Qwen TTS
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| ref_audio | AUDIO | Reference audio (ComfyUI Audio) | |
| ref_text | STRING | — | |
| model_choice | COMBO | 0.6B | 2 options: 0.6B, 1.7B |
| device | COMBO | auto | 5 options: auto, cuda, xpu, mps, cpu |
| precision | COMBO | bf16 | 2 options: bf16, fp32 |
| attention | COMBO | auto | Attention implementation |
| x_vector_onlyopt | BOOLEAN | false | If True, only speaker embedding is extracted (ref_text not needed) |
| unload_model_after_generateopt | BOOLEAN | false | Unload model from memory after generation |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| voice_clone_prompt | VOICE_CLONE_PROMPT | — |