What Is ACE-Step 1.5?
ACE-Step 1.5 is an open-weights music model built for length and control. Set the exact duration from 10 to 600 seconds, optionally pin BPM, key and time signature, and download the lossless FLAC master. Credits scale with length — a 120-second track with thinking mode runs 20, and every track starts at 10.
Free to Start
Sign up and get free credits to test every model — no subscription required.
Commercial Ready
Every song you generate is licensed for commercial use at no extra cost.
What ACE-Step Does Best
Long-form, lossless and parameter-driven.
Long-form without edits
A 600-second ceiling makes ACE-Step the model for soundtracks, mixes, meditation audio and game loops — one generation instead of stitching extend after extend.
True musical control
Pin the BPM, the key and scale, the time signature and even the vocal language — real musical parameters, not just prompt adjectives.
Lossless FLAC masters
Every ACE-Step song downloads as FLAC — the lossless original, free with the song, ready for further mixing or archival.
ACE-Step 1.5 at a Glance
Price
from 10 credits
Output
1 song per run
Access
Free tier available
Input limits
Lyrics 4,096 · Style 512 · Title 80 chars
Versions & Controls
Duration, thinking mode and advanced musical parameters.
Duration slider, 10–600s
Set the exact target length. Credits scale by the second: 0.08 per second with thinking off, 0.16 with it on, minimum 10 — the generate button always shows the live number.
Thinking mode
The default-on toggle spends extra compute planning the arrangement before generating — roughly double the credits for a more coherent long track. Turn it off for fast drafts.
Advanced musical params
BPM, key and scale, time signature (2/4 to 6/8) and vocal language — all optional. Leave them empty and the model infers them from your prompt.
Try ACE-Step 1.5 With These Prompts
One click drops the prompt straight into the editor above — tweak it, then generate.
A 300-second epic orchestral fantasy score with building strings and choir
A chill lo-fi hip hop instrumental with vinyl crackle and mellow keys, 120 seconds
A driving synthwave instrumental with arpeggiated bass and gated drums, 180 BPM
About the ACE-Step Music Model
The model
ACE-Step is an open-weights music generation project, and version 1.5 is the generation served on RaoMusic through hosted inference. Its design bets on explicit control: rather than guessing intent from a paragraph, it takes musical parameters — duration, tempo, key — as first-class inputs.
How you drive it
That makes it the long-form specialist: up to 600 seconds in one pass for soundtracks, ambient beds, meditation audio and game loops, priced by the second so a short sting costs a fraction of a long score. Simple mode has an Instrumental toggle; custom mode takes your lyrics and a style prompt.
When to choose ACE-Step
Choose it for
- Up to 600 seconds in one pass — the longest single-run ceiling on RaoMusic.
- BPM, key and time signature as real inputs, plus a free lossless FLAC.
The trade-off
No extend or cover yet — set the final length at generation time.
Better fit elsewhere
- Post-generation editing like extend and cover Suno
- A cinematic one-shot with a free WAV Google Lyria 3.5
How to Generate Music with ACE-Step
Set the length, shape the sound, download lossless.
- 1
Set length and sound
Drag the duration slider, describe the music — or write lyrics in custom mode.
- 2
Generate one FLAC track
Optionally pin BPM, key and time signature, then generate one FLAC track.
- 3
Download lossless
Download the lossless FLAC master, free with the song.
ACE-Step on RaoMusic — FAQs
Yes to both. RaoMusic serves ACE-Step 1.5 through official API partners, running hosted inference for the open-weights model, and every song carries a commercial license at no extra cost.
By the second: 0.08 credits per second with thinking mode off, 0.16 with it on, rounded up with a 10-credit minimum. The generate button shows the live cost as you move the duration slider.
From 10 to 600 seconds in a single generation — the longest single-pass ceiling of any model on RaoMusic. Set it exactly with the duration slider.
A default-on toggle that spends extra compute planning the arrangement before generating — fuller, more coherent long tracks at roughly double the credit cost. Turn it off for quick drafts.
Explore More AI Music Models
Compose to Length
Up to 600 seconds, lossless FLAC and real musical parameters — priced by the second, from 10 credits.