This is a simplified guide to an AI model called Yue maintained by Fofr. If you like these kinds of analysis, you should join AImodels.fyi or follow us on Twitter.
YuE transforms lyrics into complete songs through an innovative AI music generation system. This open-source foundation model can generate full compositions with vocals and accompaniment in multiple languages and musical styles. Unlike similar models like MusicGen which focuses on instrumental generation, YuE specializes in creating cohesive songs with lyrics.
Model Inputs and Outputs
The model takes genre descriptions and structured lyrics as input to generate complete musical compositions. It offers flexible control through genre tags and lyrical structure while maintaining musical coherence.
Inputs
- Genre Description: Text tags describing musical style, mood, vocals and instruments
- Lyrics: Text structured with verse/chorus/bridge markers
- Number of Segments: Controls song length and structure (1-10 segments)
- Max Tokens: Sets generation length (500-3000 tokens)
- Optional Audio Reference: For style matching in ICL mode
Outputs
- Complete Songs: Full compositions with vocals and accompaniment
- Multiple Audio Files: Separate vocal and instrumental tracks
- Various Formats: Generated in high-quality audio formats
Capabilities
The system can generate songs in multip...
Top comments (0)