Multilingual AI music generation nodes for ComfyUI powered by HeartMuLa.
Nodes:
- Model Loader: Load HeartMuLa 3B model with optional 4-bit quantization
- Conditioning: Tokenize lyrics and tags with popup lyrics editor
- Sampler: Generate audio tokens with CFG and temperature control
- Decode: Convert tokens to waveform using HeartCodec
- Tags Builder: Build style tags with genre/vocal/mood selection
- Transcribe: Extract lyrics from audio
Features:
- Generate up to 4 minutes of music with vocals
- Support for English, Chinese, Japanese, Korean, Spanish
- Song structure markers: [Verse], [Chorus], [Bridge], etc.
- Style control via tags (genre, vocal type, mood, tempo)
- Frontend lyrics editor with section buttons
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
