ComfyUI Node
Raon Whisper Transcribe
Transcribe reference AUDIO with Whisper for Raon-OpenTTS voice cloning. Audio output is the input passed through unchanged.
Raon Whisper Transcribe
- audio
- audio
- transcript
◄modelwhisper-large-v3-turbo►
◄dtypeauto►
◄languageauto►
◄tasktranscribe►
◄chunk_length_s30►
◄download_if_missingtrue►
CategoryRaonOpenTTS
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| audio | AUDIO | Reference audio to transcribe for the Raon-OpenTTS ref_text input. | |
| model | COMBO | whisper-large-v3-turbo | Whisper ASR model. Turbo is fast and usually accurate enough for reference transcripts. Downloads into ComfyUI/models/audio_encoders when missing. |
| dtype | COMBO | auto | Whisper precision. auto uses bf16 on supported CUDA/XPU and fp32 otherwise. |
| language | COMBO | auto | Reference audio language. auto detects it; setting it can improve transcript accuracy. |
| task | COMBO | transcribe | transcribe keeps the original language; translate outputs English. |
| chunk_length_s | INT | 300–120 | Whisper chunk length for longer reference clips. 0 lets Transformers choose. |
| download_if_missing | BOOLEAN | true | Download the selected Whisper model into ComfyUI/models/audio_encoders if it is missing. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| audio | AUDIO | — |
| transcript | STRING | — |