ComfyUI Node
VoxCPM TTS (Advanced)
A ComfyUI node in 🧪AILab/🔊TTS/VoxCPMTTS with 16 inputs and 2 outputs.
VoxCPM TTS (Advanced)
- reference_audio
- AUDIO
- STRING
◄modelVoxCPM 1.5►
◄textHello, this is VoxCPM TTS.►
◄cfg_value2.0►
◄inference_steps10►
◄max_length4096►
◄normalizetrue►
◄reference_text►
◄fade_in_ms20►
◄auto_transcribe_referencefalse►
◄show_transcription_logtrue►
◄unload_modelfalse►
◄retry_attempts2►
◄retry_threshold8.0►
◄deviceauto►
◄seed-1►
Category🧪AILab/🔊TTS/VoxCPMTTS
Inputs (16)
| Name | Type | Default | Description |
|---|---|---|---|
| model | COMBO | VoxCPM 1.5 | Select VoxCPM model version |
| text | STRING | Hello, this is VoxCPM TTS. | Text to synthesize into speech |
| cfg_value | FLOAT | 2.01–10 | Guidance scale: higher = more adherence to prompt, lower = more natural |
| inference_steps | INT | 101–100 | Number of diffusion steps: higher = better quality, lower = faster |
| max_length | INT | 4096256–8192 | Maximum token length during generation |
| normalize | BOOLEAN | true | Enable text normalization for numbers, punctuation, etc. |
| reference_audioopt | AUDIO | Reference audio for voice cloning | |
| reference_textopt | STRING | Text transcript of reference audio (leave empty for auto-transcription) | |
| fade_in_msopt | INT | 200–1000 | Fade-in duration to reduce initial artifacts (0-1000ms) |
| auto_transcribe_referenceopt | BOOLEAN | false | Automatically transcribe reference audio when no text is provided |
| show_transcription_logopt | BOOLEAN | true | Show ASR transcription logs in console |
| unload_modelopt | BOOLEAN | false | Unload model after generation to free VRAM |
| retry_attemptsopt | INT | 20–10 | Number of retry attempts for bad cases |
| retry_thresholdopt | FLOAT | 8.02–20 | Audio-to-text ratio threshold for retry detection |
| deviceopt | COMBO | auto | Device to run the model on |
| seedopt | INT | -1-1–18446744073709550000 | Random seed (-1 for random) |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| AUDIO | AUDIO | — |
| STRING | STRING | — |