GitHub survey · snapshot 2026-08-04
Best repos for music development — from audio to video
A curated path through open-source tools for building music software: analysis and DSP, native plugin frameworks, live coding and Web Audio, DAWs, generative AI, visualization, and programmatic music video.
The spectrum
Music development rarely stays in one layer. The strongest repos sit on a pipeline that starts in samples and ends in frames.
- 01AnalyzeDSP & features
- 02EngineNative & plugin
- 03ComposeLive code & Web Audio
- 04ProduceDAWs & notation
- 05GenerateAI music models
- 06See & cutViz & video
Stage picks
Each table is sorted by stars within the stage. “Why it matters” is the development angle — not consumer features.
01 · Analyze — DSP & features
| Repo | Stars | Lang | Why it matters |
|---|---|---|---|
| librosa/librosa | 8.5k | Python | Default toolkit for spectral features, onset, beat, and MIR prototypes. |
| spotify/pedalboard | 6.2k | C++ / Python | Production-grade effects chain from Python; bridges research and plugins. |
| MTG/essentia | 3.7k | C++ | Industrial MIR library with Python bindings — descriptors beyond hobby DSP. |
| olilarkin/awesome-musicdsp | 3.0k | — | Best curated map of DSP papers, SDKs, and plugin resources. |
02 · Engine — native frameworks & plugin APIs
| Repo | Stars | Lang | Why it matters |
|---|---|---|---|
| AudioKit/AudioKit | 11.4k | Swift | Apple-platform synthesis, analysis, and processing stack. |
| juce-framework/JUCE | 8.7k | C++ | Industry standard for VST3 / AU / AAX / LV2 hosts and plugins. |
| FluidSynth/fluidsynth | 2.4k | C | SoundFont synth used everywhere MIDI needs a reliable voice. |
| iPlug2/iPlug2 | 2.4k | C++ | Plugin framework covering desktop, mobile, XR, and web targets. |
| free-audio/clap | 2.3k | C | Modern open plugin API — the post-VST conversation that matters. |
| Tracktion/tracktion_engine | 1.4k | C++ | Embeddable DAW engine if you are shipping a host, not just an effect. |
03 · Compose — live coding & Web Audio
| Repo | Stars | Lang | Why it matters |
|---|---|---|---|
| goldfire/howler.js | 25.3k | JavaScript | Battle-tested web audio playback abstraction for products. |
| Tonejs/Tone.js | 14.7k | TypeScript | The Web Audio composition library — scheduling, synths, transport. |
| sonic-pi-net/sonic-pi | 11.9k | C++ | Live-coding instrument that taught a generation to code music. |
| katspaugh/wavesurfer.js | 10.4k | TypeScript | Waveform UI primitive for editors, players, and annotation tools. |
| supercollider/supercollider | 6.7k | C++ | Server + language for algorithmic synthesis and research. |
| overtone/overtone | 6.2k | Clojure | Programmable music on SuperCollider — live coding with teeth. |
| scribbletune/scribbletune | 3.8k | TypeScript | JS-first sequencing and MIDI generation for browser pipelines. |
| chaosprint/glicol | 3.0k | Rust | Graph-oriented live coding / DSP — WASM-friendly Rust audio. |
04 · Produce — DAWs & notation
| Repo | Stars | Lang | Why it matters |
|---|---|---|---|
| musescore/MuseScore | 14.9k | C++ | Reference open notation app — MusicXML, engraving, education. |
| LMMS/lmms | 10.2k | C++ | Full open DAW; strong for studying sequencer and plugin hosting UX. |
| mixxxdj/mixxx | 7.0k | C++ | DJ performance stack — decks, sync, controllers, real-time constraints. |
| Ardour/ardour | 5.2k | C++ | Pro-grade open DAW; deep reference for session and routing design. |
| andremichelle/openDAW | 2.0k | TypeScript | Actively shipping next-gen browser DAW — watch this space. |
| gridsound/daw | 1.8k | JavaScript | Mature Web Audio DAW effort with long public iteration history. |
| DBraun/DawDreamer | 1.3k | C++ | Headless DAW for Python — VST, FAUST, JAX, offline render pipelines. |
05 · Generate — AI music & separation
| Repo | Stars | Lang | Why it matters |
|---|---|---|---|
| Anjok07/ultimatevocalremovergui | 25.6k | Python | Stem separation GUI that made DNN source separation mainstream. |
| facebookresearch/audiocraft | 23.5k | Python | MusicGen + EnCodec — still the clearest open text-to-music baseline. |
| magenta/magenta | 19.8k | Python | Long-running Google research stack for symbolic and audio generation. |
| ace-step/ACE-Step-1.5 | 12.0k | Python | Hottest local music-gen model family in 2026 — Mac/AMD/CUDA. |
| facebookresearch/demucs | 10.4k | Python | Hybrid spectrogram/waveform separation — research code still cited daily. |
| open-mmlab/Amphion | 10.2k | Python | Unified toolkit for audio, music, and speech generation research. |
| multimodal-art-projection/YuE | 6.4k | Python | Open full-song generation model — Suno-class ambition, open weights. |
| Stability-AI/stable-audio-tools | 3.8k | Python | Conditional audio generation tooling from Stability. |
| HeartMuLa/heartlib | 3.8k | Python | 2026 open music-gen contender — rising fast with studio frontends. |
| 0xShug0/audio.cpp | 1.1k | C++ | ggml-powered C++ inference for TTS/STT/music — no Python required. |
06 · See & cut — visualization and music video
| Repo | Stars | Lang | Why it matters |
|---|---|---|---|
| remotion-dev/remotion | 55.4k | TypeScript | Programmatic video with React — the music-video render path for web teams. |
| openframeworks/openFrameworks | 10.4k | C++ | Creative-coding toolkit for live AV installs and reactive visuals. |
| projectM-visualizer/projectm | 4.4k | C++ | Milkdrop-compatible music visualization library, cross-platform. |
| hydra-synth/hydra | 2.7k | JavaScript | Browser livecoding visuals — networked, algorave-adjacent. |
| jberg/butterchurn | 1.9k | JavaScript | WebGL Milkdrop — drop-in visualizer for web players and streams. |
| hvianna/audioMotion-analyzer | 0.9k | JavaScript | High-res spectrum analyzer module — zero deps, product-ready. |
Trending watchlist
High-velocity repos (created or spiked since 2024) that are shaping the next tools — not always the highest stars overall.
- ACE-Step 1.5 — local text-to-music with commercial-grade hype; UI forks and DAW shells already circling it.
- openDAW — actively pushed browser DAW; strongest signal that Web Audio production apps are real.
- YuE — open full-song generation; the “Suno but open” research track.
- HeartMuLa / heartlib — 2026 model release with studio frontends appearing within weeks.
- audio.cpp — ggml C++ inference for audio models without a Python runtime.
- CLAP — plugin API momentum; pair with JUCE/iPlug2 if you ship instruments.
Method
- Queried GitHub Search (stars + topics:
web-audio,daw,music-generation, and related terms) on 2026-08-04. - Favored development surfaces: engines, APIs, DAWs, generation toolkits, viz/video frameworks.
- Excluded pure streaming clients, downloaders, and piracy lists even when star counts were higher.
- Star counts rounded for readability; treat them as relative signal, not a ranking algorithm.