by ACE Studio
Open-source music generation with fine-grained control over BPM, key, lyrics, and instrumentation.
ACE-Step is an open-source music generation foundation model by ACE Studio that synthesizes full songs from text descriptions. It integrates diffusion-based generation with deep compression, achieving state-of-the-art speed — up to 4 minutes of music in 20 seconds on an A100 GPU. ACE-Step v1.5 runs locally with less than 4GB VRAM.
Combine genre, mood, and instrument descriptors:
lo-fi hip-hop, chill, nostalgic, piano, soft drums,
vinyl crackle, rainy day vibes, 85 bpm
For precise control, include tempo and musical key:
synthwave, energetic, 128 bpm, key of Am,
pulsing synth bass, arpeggiated leads, 80s drums
Extreme BPM values have less training data and may produce unstable results.
[Verse] and [Chorus] tags may not work as expected. Focus on genre tags, mood, instruments, and production descriptors instead.
| Parameter | Description |
|---|---|
| bpm | Tempo (stable: 60-180) |
| key | Musical key (C, G, D, Am, Em most stable) |
| genre | pop, rock, electronic, hip-hop, jazz, etc. |
| mood | happy, sad, energetic, chill, dark, epic |
| instruments | piano, guitar, drums, bass, synth, strings |
| vocals | male vocal, female vocal, choir, no vocals |
| instrumental | Boolean — no vocals |
| duration | Duration in seconds |
| time_signature | 4/4, 3/4, 6/8, etc. |
Quick tips from the community about what works with ACE-Step right now.
Sign in to share a tip.
No tips yet. Add a tip for this model.