The AI-native DAW where every instrument knows what came before it.
Quick Start • What Makes It Different • Features • Architecture • Contributing
Want a production-ready experience today? Try ACE Studio — our professional AI music creation platform with higher quality models and real-time collaboration.
Most AI music tools generate a full song in one shot. ACE-Step DAW takes a fundamentally different approach — it builds songs layer by layer, where each new instrument is musically aware of everything generated before it.
Generate drums first. Then bass that grooves with the drums. Then guitar that complements both. Then vocals that sit perfectly in the mix. Each layer "hears" the cumulative result of all previous layers through the ACE-Step 1.5 model.
Drums → Bass (hears drums) → Guitar (hears drums+bass) → Vocals (hears everything)
This isn't post-hoc mixing — the AI composes each part knowing the full musical context, producing arrangements that are coherent from the ground up.
Write s("[bd <hh oh>]*2, [~ cp]*2") and hear it instantly. ACE-Step DAW integrates Strudel as a first-class track type alongside traditional timeline clips. Switch between visual editing and code-based pattern generation without leaving the DAW. Convert between MIDI and Strudel patterns freely.
Every feature is accessible via window.__store.getState().actionName(). The entire DAW state is scriptable, testable, and automatable. Built for a future where AI agents and humans collaborate on music production side by side.
Requirements: Node.js 18+ and an ACE-Step 1.5 API server (local or cloud).
git clone https://github.com/ace-step/ACE-Step-DAW.git
cd ACE-Step-DAW
npm install
npm run devOpens at http://localhost:5174. The dev server proxies /api to the ACE-Step 1.5 backend at localhost:8001.
Cloud API: Configure the backend URL in Settings → API URL: https://api.acemusic.ai
Production build:
npm run build && npm run preview| Feature | Description |
|---|---|
| LEGO Pipeline | Sequential generation with cumulative context — drums → bass → guitar → vocals |
| Text-to-Music | Describe what you want, get audio — 16 genre presets (Pop, Rock, Jazz, Electronic, Hip-Hop, Classical, Lo-Fi, Ambient) |
| Cover Generation | Reimagine existing audio in a new style with controllable deviation strength |
| Repaint | Selectively regenerate a time range — conservative, balanced, or aggressive modes |
| Vocal2BGM | Feed in vocals, generate matching accompaniment automatically |
| Audio Analysis | AI-powered BPM, key, and genre detection |
| Model Selector | Choose DiT/LM models, adjust inference steps, LoRA support |
| Feature | Description |
|---|---|
| Multi-Track Timeline | Clip-based arrangement with drag, resize, split, fade handles, and snap-to-grid |
| 5 Track Types | Stems (AI), Sample (imported audio), Sequencer (step drums), Piano Roll (MIDI), Strudel (live code) |
| Piano Roll | Canvas-based MIDI editor with velocity lane, slide notes, ghost notes, chord stamps, and quantize |
| Step Sequencer | FL Studio-inspired 16-step drum pattern editor with 16 synthesized drum sounds and 4 kit presets |
| Effect Chain | Per-track effect rack — EQ3, Compressor, Reverb, Delay, Distortion, Chorus, Flanger, Phaser, Filter, Sidechain |
| Mixer | Channel strips with volume, pan, mute, solo, and per-track metering |
| 6 Synth Presets | Piano, Strings, Pad, Lead, Bass, Organ (Tone.js polyphonic synthesis) |
| Loop Browser | 15 built-in synthesized loops with search, filter, and drag-to-timeline |
| Recording | Microphone input with real-time waveform, count-in, and input level metering |
| Automation | Breakpoint envelopes with interpolation for volume, pan, and effect parameters |
| Project Persistence | Save/load via IndexedDB with auto-save |
| WAV Export | Offline render to stereo WAV |
| Undo/Redo | 55+ undoable actions covering virtually every operation |
| 25+ Keyboard Shortcuts | Full keyboard-driven workflow — see full list |
┌──────────────────────────────────────────────────────────────┐
│ React 19 Components (60+) │
│ Timeline │ PianoRoll │ Sequencer │ Mixer │ Strudel Editor │
└──────────────────────────┬───────────────────────────────────┘
│
┌──────────────────────────▼───────────────────────────────────┐
│ Zustand Stores (7) │
│ project │ transport │ generation │ ui │ model │ shortcuts │
└──────────────────────────┬───────────────────────────────────┘
│
┌──────────────────────────▼───────────────────────────────────┐
│ Services Layer │
│ Generation Pipeline │ ACE-Step API │ Audio File Manager │
│ Strudel Conversion │ Project Storage │
└──────────────────────────┬───────────────────────────────────┘
│
┌──────────────────────────▼───────────────────────────────────┐
│ Audio Engine (Tone.js + Web Audio API) │
│ AudioEngine │ SynthEngine │ DrumEngine │ EffectsEngine │
│ AutomationEngine │ RecordingEngine │ StrudelEngine │
└──────────────────────────┬───────────────────────────────────┘
│
┌─────▼─────┐
│ Speaker / │
│ WAV File │
└───────────┘
Signal chain per track:
Source → Input Gain → Pan → EQ3 → Effect Chain → Compressor → Volume → Analyser → Master
src/
components/ # 60+ React components by domain
timeline/ # Arrangement view, clip blocks, track lanes
pianoroll/ # Canvas MIDI editor, velocity lane
sequencer/ # Step grid, beat pads
mixer/ # Channel strips, effect chain
strudel/ # Live code editor
generation/ # AI generation panels
transport/ # Transport bar, tempo/time
engine/ # Audio engine (Tone.js + Web Audio wrappers)
services/ # Business logic (API, generation pipeline, storage)
store/ # Zustand stores
hooks/ # React hooks (transport, recording, shortcuts)
types/ # TypeScript interfaces
utils/ # Pure helpers (WAV, waveform, color, time)
constants/ # Defaults, presets, track definitions
| Layer | Technology |
|---|---|
| UI | React 19 + TypeScript 5.7 + Vite 6 |
| Styling | Tailwind CSS v4 |
| State | Zustand 5 |
| Audio | Tone.js 15 + Web Audio API |
| Live Coding | Strudel (core, mini, webaudio, soundfonts) |
| Storage | IndexedDB via idb-keyval |
| Testing | Vitest + Playwright |
| Key | Action | Key | Action | |
|---|---|---|---|---|
Space |
Play / Pause | X |
Toggle mixer | |
Enter |
Stop | B |
Toggle smart controls | |
R |
Record | O |
Toggle loop browser | |
L |
Toggle loop | E |
Edit selected clip | |
K |
Toggle metronome | S |
Split clip at playhead | |
N |
Toggle snap | Q |
Quantize notes | |
Z |
Zoom to fit | Delete |
Delete selection | |
Home/End |
Seek start/end | Cmd+Z |
Undo | |
←/→ |
Nudge ±5s | Cmd+Shift+Z |
Redo | |
Cmd+Scroll |
Zoom timeline | Cmd+D |
Duplicate clip | |
Cmd+Shift+I |
Add track | Cmd+Shift+E |
Export WAV | |
? |
Show all shortcuts | Cmd+A |
Select all |
npm run dev # Dev server at http://127.0.0.1:5174
npm test # Unit tests (Vitest)
npm run test:e2e # E2E tests (Playwright)
npm run test:all # Unit + E2E
npm run test:coverage # Coverage report
npm run build # Type-check + production build| Metric | Value |
|---|---|
| Source code | ~24,000 LOC |
| Components | 60+ |
| Unit tests | 94 |
| E2E tests | 21 |
| CI pipeline | type-check → unit-test → build → e2e → review |
See AGENTS.md for the complete development workflow and agent conventions.
We welcome contributions! Whether it's a bug fix, new feature, or documentation improvement:
- Find or create an issue — check open issues or create one with a
feat:,fix:, ordocs:prefix - Fork and branch —
feat/issue-NUMBERorfix/issue-NUMBER - Write tests first — we follow TDD (red → green → refactor)
- Pass quality gates —
npx tsc --noEmit+npm test+npm run buildmust all succeed - Open a PR — include
Closes #NUMBERin the body
See AGENTS.md for detailed conventions on commits, testing, and code style.
AGPL-3.0-or-later — ACE-Step DAW bundles Strudel packages (AGPL-3.0-or-later), so the project is distributed under AGPL-compatible terms as a whole.
If you need an MIT-only distribution, remove the Strudel integration or obtain an alternative license from the Strudel copyright holders. See THIRD_PARTY_NOTICES.md for details.
For hosted deployments, make source code available to network users. Override branding with:
VITE_SOURCE_CODE_URL, VITE_LICENSE_URL, VITE_COPYRIGHT_NOTICE
ACE-Step DAW is an open-source project by ACE Studio.
For professional production with higher quality models and real-time collaboration, visit acestudio.ai.