Wavy Labs — Free, Open-Source AI-Powered DAW

"The DAW that listens to you."

Wavy Labs is a free, open-source AI-powered Digital Audio Workstation forked from LMMS (GPL-2.0). It layers powerful AI features on top of a full-featured desktop DAW — powered by ElevenLabs, Anthropic Claude, and Groq, with local models for offline use. You bring your own API keys.

---

Download

Download for Windows (v0.15.0)

System requirements: Windows 10/11 (64-bit), 8 GB RAM, 4 GB disk space. Optional: NVIDIA GPU for GPU-accelerated stem splitting and local models.

---

Features

• Full DAW — complete LMMS fork with MIDI piano roll, Song editor, pattern editor, 30+ synth plugins
• AI Music Generation — ElevenLabs cloud or local DiffRhythm/ACE-Step models
• Stem Splitting — Demucs v4 (2/4/6-stem) on any audio
• TTS / SFX / Voice Clone — ElevenLabs voice tools
• Vocal Processing — voice isolation, transcription, forced alignment, dubbing
• Prompt Commands — natural-language DAW control via Claude / Groq
• AI Mix & Master — automatic mix analysis and mastering
• Instrument Library — 570+ searchable presets, samples, GM patches, VST3 references
• Genre Modes — 8 genre presets with automatic instrument + drum kit configuration
• Code to Music — generate audio/MIDI from a Python DSL

All features are included — no subscription required. Add your own API keys to enable cloud AI.

---

Quick Start

1. Download & Install

Download the installer from Releases and run it. The installer bundles an embedded Python runtime and installs all dependencies automatically.

2. Configure API Keys

On first launch, the setup wizard will prompt for API keys. You can also add them later via Edit → Settings:

Key	Purpose	Get it at
Anthropic	Claude — prompt commands, chat	console.anthropic.com
Groq	llama-3.3-70b — free LLM fallback	console.groq.com/keys
ElevenLabs	Music, TTS, SFX, voice tools	elevenlabs.io/app/settings/api-keys
Freesound	Browse/download sound effects	freesound.org/apiv2/apply

Keys are stored encrypted on your machine (Windows DPAPI) and never sent to Wavy Labs.

---

Build from Source

Prerequisites

• Qt 6.9.x (with Quick, QuickControls2, Svg, Sql, Network)
• MSVC 2022 (Windows) or GCC 13+ (Linux)
• CMake 3.22+, Ninja
• Python 3.10+ with pip
• ZeroMQ 4.x + cppzmq

Windows (MSVC)

git clone https://github.com/stroland02/Wavy-Labs-Code-Base-.git WL
cd WL
git submodule update --init --recursive

# Configure (edit paths in do_configure.bat first)
do_configure.bat

# Build
do_build_lmms.bat

# Run
build\lmms.exe

Python backend

cd wavy-ai
pip install -r requirements.txt

# Optional: GPU-accelerated torch (NVIDIA RTX)
pip install torch torchaudio --index-url https://download.pytorch.org/whl/cu126

python server.py

The C++ app will launch the Python backend automatically. For development, start it manually first — the app detects an existing backend and skips auto-launch.

---

Repository Structure

WL/
├── lmms-core/          # Git submodule: forked LMMS (Qt6, branch wavy-patches)
├── wavy-ui/            # C++/Qt6 panels (static lib wavy_ui)
│   ├── AIPanel/        # Legacy AI generation panel
│   ├── QML/            # AIBackend + QML bindings (MainPanel.qml)
│   ├── Shell/          # WavyShell — top-level window replacing LMMS MDI
│   ├── Transport/      # Transport bar + genre config popup
│   ├── LicenseGate/    # LicenseManager (tier=Studio always), ApiKeySettings
│   └── IPC/            # ZeroMQ JSON-RPC client (AIClient) + BackendLauncher
├── wavy-ai/            # Python AI backend (ZeroMQ JSON-RPC server)
│   ├── server.py       # Entry point
│   ├── rpc_handlers.py # Method → handler registry
│   ├── rpc/            # Domain handlers (core, audio, midi, library, fx, …)
│   ├── cloud/          # Cloud provider wrappers (ElevenLabs, Anthropic, Groq)
│   ├── models/         # Local model wrappers (Demucs, DiffRhythm, ACE-Step)
│   └── data/           # instrument_catalog.json, soundfonts
├── wavy-installer/     # NSIS Windows installer
├── wavy-license-server/# Optional: open-source license server (FastAPI)
├── data/               # QML, QSS themes, icons, resources.qrc
└── src/main.cpp        # Wavy entry point (bypasses lmms-core main.cpp)

---

IPC Protocol

C++ (AIClient)  ──JSON-RPC over ZeroMQ──►  Python (server.py)
                   tcp://127.0.0.1:5555

All methods use: {"id": N, "method": "method_name", "params": {...}}

Key RPC methods: health, generate_music, split_stems, prompt_command, compose, elevenlabs_tts, elevenlabs_sfx, elevenlabs_voice_clone, update_api_keys, list_instruments, browse_dataset, and 40+ more.

---

Contributing

1. Fork the repo and create a branch
2. Make your changes (follow existing code patterns)
3. Test: do_build_lmms.bat → build\lmms.exe
4. Open a pull request

Bug reports and feature requests: GitHub Issues

---

License

• DAW core (lmms-core/): GPL-2.0 (same as LMMS upstream)
• wavy-ui/, wavy-ai/, wavy-license-server/: GPL-2.0
• AI model weights: see individual model licenses (Demucs: MIT, DiffRhythm: Apache 2.0, etc.)
• Instrument samples: see wavy-ai/data/instrument_catalog.json for per-entry licenses