The same open ACE-Step model you'd self-host — without the months of pipeline work it takes to make it release-ready.
Run text-to-music generation from a single REST call. Send studio_quality: true and get production-grade audio back — vocals, instrumentation and mix already handled. No GPUs to provision, no post-processing chain to build.
Built on the open-source ACE-Step model · REST · async tasks · royalty-free output
# Generate a release-ready track curl -X POST https://acestep.io/api/v2/generate-audio \ -H "Authorization: Bearer sk-your-api-key-here" \ -H "Content-Type: application/json" \ -d ' "tags": "pop, energetic, synth, dance", "lyrics": "[Verse 1]\n...", "seconds": 60, "steps": 12, "studio_quality": true ' # → 1 studio task, 10 credits "code": 0, "data": "tasks": [ "task_id": "83082478890054944" ], "credits_consumed": 10, "remaining_credits": 771
# Poll until the track is ready (free) curl https://acestep.io/api/v2/task-status/83082478890054944 \ -H "Authorization: Bearer sk-your-api-key-here" # → when status is "completed" "code": 0, "data": "task_id": "83082478890054944", "status": "completed", "progress": 100, "audio_url": "https://cdn.acestep.io/...mp3", "lrc": "[00:00.00]...", "created_at": "2026-06-05T15:54:33Z"
Getting ACE-Step running is easy. Getting it release-ready is the hard part.
ACE-Step 1.5 is open source — the weights are right there, and spinning up inference is a weekend project. The wall comes after. Raw model output lands somewhere short of production: vocals that sit too far back, arrangements that drift instead of locking into a groove, results that hold up in one genre and fall apart in another.
Closing that gap means building and tuning an automated post-processing and calling layer — and that's the part that quietly takes months.
One team came to us after exactly this. They'd set out to self-host ACE-Step on their own GPUs, but the generation quality fell short of what they could ship.
# the gap is why this page exists →What studio_quality: true actually does
It's the same foundation model underneath. What we wrap around it is the layer you'd otherwise build yourself.
Clearer, more present vocals
Automated post-processing brings the vocal forward and cleans it up, so it reads as a lead — not a layer buried in the mix.
Tighter rhythmic feel
Generation presets tuned for a steady, locked-in groove instead of arrangements that wander off the beat.
Holds across genres
Pop, rock, ballad, rap — the vocal treatment adapts instead of only working on one kind of track.
One flag, studio_quality: true, routes your request through that whole layer. You get back a single release-ready track instead of raw output you'd need to finish yourself.
From one call to release-ready audio
Asynchronous by design — built for server-side and batch generation, not just one-off requests.
Generate
One POST with your tags, lyrics and studio_quality: true. Returns a task ID right away.
Poll & fetch
Poll task-status until it reads completed, then pull the audio_url.
Generation returns a task_id; you poll task-status until completed, then download from audio_url. That's the whole integration — see the full docs →
Already running, every day
A team building server-side batch generation switched to the API after trying to self-host. It's been running continuously on studio_quality: true ever since.
They didn't need a better model — they had the same one. They needed it release-ready without building the pipeline.
Built for teams putting music inside their product
If music generation is a feature you ship — not a one-off you make by hand — this is the integration path.
Apps
Background scores and soundtracks generated on demand, per user or per session.
Games
Adaptive or batch-generated audio that fits the scene and the moment.
Content pipelines
Original, royalty-free tracks at volume, with no clearance overhead.
Agents & DApps
Programmatic music as a building block inside a larger system.
Teams building music into apps, games and content products are integrating the same API.
A full production toolchain, not just generation
The same API extends tracks, replaces sections, adds stems, builds covers and writes lyrics.
API access is part of the Studio plan
API key creation is available on the Studio plan. You can read the full docs publicly before you commit — keys and live calls require an active Studio subscription.
ACE-Step API, answered
Yes — ACE-Step is open source and you're free to run it yourself. The API exists for teams that don't want to build the release-ready layer on top: the automated post-processing and calling layer that turns raw model output into production-grade audio.
It routes your request through the post-processing and tuning layer for production-grade output — clearer vocals, a tighter rhythmic feel and treatment that holds across genres. A standard request consumes 5 credits; a Hi-Fi request consumes 10.
Yes. We run the ACE-Step model and expose it over REST at https://acestep.io/api/v2/, so you don't have to provision GPUs or manage inference. Authenticate with a Bearer token and call it from anywhere.
No — it's asynchronous. You create a task, poll task-status until it reads completed, then fetch the audio_url. This design fits server-side and batch generation cleanly.
No. AceStep is an independent product built on the open-source ACE-Step model. It is not affiliated with, endorsed by or connected to Suno.
Yes. Output on paid plans is royalty-free and cleared for commercial use, so you can ship generated music inside your product.
Ship music your users can actually release.
Read the docs, then start calling. studio_quality: true does the finishing for you.