MusicGen Large
MetaAudio & Speech
82.0
Performance
★ 4.2
Rating
340
Reviews
Music GenerationOpen SourceMedium
About
Meta's open-source music generation model creating high-quality music from text descriptions or melody inputs.
Strengths
Strong open-source music generation from text prompts. Melody-conditioned generation lets you hum a tune and get a full arrangement. Multiple genres and styles. Stereo audio output. Part of Meta's AudioCraft framework.
Specifications
- Context window
- —
- Parameters
- 3.3B
Available On
HuggingFaceReplicateLocal deployment
Features
text to musicmelody conditioningstereo audiomulti genre
Performance Trend
Benchmark score trends over time for the top 5 benchmarks.
Loading history...
Benchmarks
Scores from various benchmark tests; higher is better.
| Test | Score | Percentile | Source |
|---|---|---|---|
| MOS | 80.0 | p78 | seed |
