List your company

Text-to-Music Generation on 100 Official Final Prompts (test)

0.531ms-CLAP ScoreMusicGen-small
0.4694320.4854160.50140.517384May 20, 2026

Evaluation results

MethodLinks
MusicGen-smallParams=300M, Data=∼20kh2026.050.5316.867.475.227.47.48.368.935.585.447.053.63.553.55
MusicGen-largeParams=3.3B, Data=∼20kh2026.050.5317.127.695.427.587.388.48.955.955.727.433.853.753.55
MusicGen-mediumParams=1.5B, Data=∼20kh2026.050.5286.857.535.257.447.428.428.925.855.727.23.853.753.8
Stable Audio OpenParams=1.2B, Data=∼7.3kh2026.050.5076.457.544.337.637.348.318.894.244.964.883.63.853.35
Efficiency: without lyric encoder (inference)Params=499M, Data=457h2026.050.50056.667.394.817.267.358.388.883.073.123.83.153.32.9
Efficiency: without timbre encoder (inference)Params=499M, Data=457h2026.050.49966.817.424.967.337.398.378.953.073.013.862.952.92.75
Efficiency Original, 16-layer DiTParams=499M, Data=457h2026.050.49936.927.45.177.337.418.48.893.082.942.953.12.75
retrained without lyric & timbre encoders, 23-layer DiTParams=489M, Data=457h2026.050.48446.57.115.027.057.358.38.922.62.63.282.752.92.65
Performance: without timbre encoder (inference)Params=2.4B, Data=457h2026.050.48266.537.314.77.257.378.378.93.553.724.323.153.253.25
retrained without lyric & timbre encoders, 24-layer DiTParams=1.8B, Data=457h2026.050.47766.477.065.27.017.48.398.952.0322.812.252.552.25
Performance: without lyric encoder (inference)Params=2.4B, Data=457h2026.050.47396.387.244.67.157.358.358.873.443.634.363.13.43.1
Performance Original, 24-layer DiTParams=2.4B, Data=457h2026.050.47376.657.374.897.297.378.358.873.343.434.252.953.22.85
retrained without lyric & timbre encoders, 16-layer DiTParams=371M, Data=457h2026.050.47186.457.035.157.027.388.338.911.941.972.512.452.52.15