4.1 KiB
4.1 KiB
FL Song Gen
AI-powered song generation nodes for ComfyUI based on Tencent's SongGeneration (LeVo) model. Generate complete songs with vocals and instrumentals from lyrics!
Features
- Full Song Generation - Generate complete songs with vocals and instrumentals
- Dual-Track Output - Get separate vocal and background music tracks
- Lyrics-to-Song - Structured lyrics with sections (verse, chorus, bridge, intro, outro)
- Style Transfer - Use reference audio to guide the musical style
- Text Descriptions - Control gender, timbre, genre, emotion, and BPM
- Auto Style Presets - Quick generation with Pop, Rock, Jazz, and more
- Long-Form Generation - Up to 4 minutes 30 seconds per song
Nodes
| Node | Description |
|---|---|
| Model Loader | Load SongGeneration model with memory options |
| Lyrics Formatter | Build properly formatted lyrics from sections |
| Description Builder | Create style descriptions from components |
| Generate | Main generation with text conditioning |
| Style Transfer | Generate using reference audio for style |
| Auto Style | Generate with preset style prompts |
Installation
Prerequisites
- Clone the SongGeneration repository to your ComfyUI parent directory:
cd /path/to/ComfyUI/..
git clone https://github.com/AslpLab/SongGeneration.git
- Download model weights from HuggingFace:
# Download to ComfyUI/models/songgen/songgeneration_base_new/
# Required files: config.yaml, model.pt
ComfyUI Manager
Search for "FL Song Gen" and install.
Manual
cd ComfyUI/custom_nodes
git clone https://github.com/filliptm/ComfyUI_FL-SongGen.git
cd ComfyUI_FL-SongGen
pip install -r requirements.txt
Quick Start
- Add FL Song Gen Model Loader and select model variant
- Add FL Song Gen Lyrics Formatter to build your lyrics
- Add FL Song Gen Description Builder for style (optional)
- Connect to FL Song Gen Generate node
- Connect outputs to audio save/preview nodes
Models
| Model | Max Duration | VRAM | Languages |
|---|---|---|---|
| songgeneration_base | 2m30s | 10-16GB | Chinese |
| songgeneration_base_new | 2m30s | 10-16GB | Chinese, English |
| songgeneration_base_full | 4m30s | 12-18GB | Chinese, English |
| songgeneration_large | 4m30s | 22-28GB | Chinese, English |
Models must be downloaded manually from HuggingFace to ComfyUI/models/songgen/.
Lyrics Format
Lyrics use section tags separated by ; with phrases separated by .:
[intro-short] ; [verse] First line.Second line.Third line ; [chorus] Chorus line one.Chorus line two ; [outro-short]
Available Section Tags:
[intro-short],[intro-medium],[intro-long]- Instrumental intro[verse]- Verse section (requires lyrics)[chorus]- Chorus section (requires lyrics)[bridge]- Bridge section (requires lyrics)[inst-short],[inst-medium],[inst-long]- Instrumental break[outro-short],[outro-medium],[outro-long]- Instrumental outro
Style Descriptions
Format: "voice_type, timbre, genre, emotion, instruments, the bpm is X"
Example: "female, warm, pop, emotional, piano and drums, the bpm is 120"
Auto Style Presets
- Pop, R&B, Dance, Jazz, Folk, Rock
- Chinese Style, Chinese Tradition, Chinese Opera
- Metal, Reggae, Auto
Requirements
- Python 3.10+
- CUDA 11.8+ (for GPU acceleration)
- 16GB RAM minimum (32GB+ recommended)
- NVIDIA GPU with 10-28GB VRAM (depends on model)
Note: CPU-only mode is supported but very slow. Mac MPS may have limited support.
License
Apache 2.0
Credits
Based on SongGeneration (LeVo) by Tencent AI Lab.