ComfyUI Node
MIDI Transcribe Audio
Transcribe audio to MIDI JSON using SoulX-Singer. Runs full preprocessing: vocal separation, F0 extraction, VAD, lyrics transcription, and note transcription. Output MIDI JSON can be edited with MIDI Edit Lyrics / MIDI Lyrics Alignment, then fed into MIDI Synthesize Audio for singing voice synthesis.
MIDI Transcribe Audio
- audio
- midi_json
◄max_merge_duration30000►
◄languageMandarin►
CategoryMIDI-SoulX
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| audio | AUDIO | — | |
| max_merge_durationopt | INT | 300001000–120000 | — |
| languageopt | COMBO | Mandarin | 3 options: Mandarin, English, Cantonese |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| midi_json | STRING | — |