AudioMIX

AI-assisted music production and live performance — one platform for the studio and the stage.

Write a script. Make a beat. Light up a room.

AudioMIX is an open-source DAW and live-performance engine for independent artists and developer-musicians. It combines a real-time C++ DSP stack, an AI creative layer (Juniper2.0), and AudioScript — a domain-specific language for controlling audio, mood, lighting, and show logic in code.

This repository is the core engine. The desktop UI lives in a separate repo: AudioMIX-Electron.

---

Table of contents

• Why AudioMIX
• Quick start
• Core features
• Architecture
• Project layout
• AudioScript
• Juniper2.0
• External services
• AudioMIX-Electron
• Development
• Ethical AI
• License

---

Why AudioMIX

Most DAWs are built for the studio. Most live-coding tools are not built for performers. AudioMIX bridges that gap.

You are…	AudioMIX gives you…
Producer	DSP, mood/ML analysis, EQ tooling, and AI-assisted production workflows
Live performer	AudioScript to script audio, lighting, and transitions in real time
Developer-musician	Modular Python/C++ extension points, a compiler pipeline, and an event-driven runtime

---

Quick start

Prerequisites

• Python 3.11 (recommended; see requirements.txt — librosa targets ≤ 3.11)
• CMake 3.10+
• System audio: PortAudio and/or PulseAudio, ALSA (Linux)
• Build tools: build-essential, ffmpeg

Optional: run ./install.sh on Debian/Ubuntu for system packages and a local amenv virtualenv.

Install and build

git clone https://github.com/alexisvassquez/audiomix.git
cd audiomix

python3 -m venv amenv && source amenv/bin/activate
pip install -r requirements.txt

cmake -S . -B build
cmake --build build -j"$(nproc)"

Verify the DSP build:

./build/test_sine_dsp

Run the AudioScript shell

python audioscript_runtime.py

Run a script file:

python audioscript_runtime.py path/to/show.audioscript
# or
python audioscript_runtime.py path/to/show.as

Useful flags:

Flag	Effect
--safe	Restrict loaded modules; no hardware/audio side effects
--no-dsp	Skip launching the C++ audiomix binary
--debug	Verbose module loading
--no-emoji	Plain text output

Set AUDIOMIX_SAFE=1 in the environment for the same allowlist as --safe.

Compile audio → AudioScript (IR)

python audioscript/compiler/run.py --file your_track.wav
python audioscript/compiler/run.py --file your_track.wav --save

---

Core features

• Real-time DSP — C++ engine with EQ, compression, gain, and more (audio/dsp/)
• AudioScript — Live shell + IR compiler for scripted performances (audioscript/)
• AI / ML analysis — Mood tagging, MFCC/spectral features, tempo and beat tracking (audio/ai/)
• Performance engine — Modular command registry, session state, event bus, DSP bridge (performance_engine/)
• Streaming integrations — Official Spotify Web API; Bandcamp embed support (audio_providers/)
• LED & lighting — Mood/BPM-driven patterns and reactive output (audio/led/, audio/led-service/)
• Audience listener — Crowd-reactive performance hooks (in development)

---

Architecture

AudioMIX is organized around two creative modes:

Mode	Context	Role of AudioScript
Studio	Desk, iteration, AI as collaborator	Author and refine scripts; use the compiler and analysis pipeline
Performance	Stage, low latency, crowd-reactive	Execute Live scripts; runtime mutates mood, DSP, and lighting via the event bus

Runtime data flow (simplified):

AudioScript shell  →  command modules  →  EventBus  →  DSPBridge (NDJSON)  →  C++ audiomix
                              ↓
                        audio/ai, LED, providers

Full directory snapshot: docs/AudioMIX_project_tree.txt

---

Project layout

Path	Purpose
audio/dsp/	C++ real-time DSP modules and tests
audio/ai/	ML inference, feature extraction, datasets, planners
audio/eq/	EQ engine, presets, live streaming
audio/led/	LED controller, color profiles, reactive logic
audio/led-service/	FastAPI microservice for mood/BPM JSON input
performance_engine/	Live runtime: modules, event bus, DSP bridge, session state
audioscript/	Language spec, EBNF grammars, compiler, example scripts
audio_providers/	Spotify and Bandcamp adapters
runtime/	Refactored shell components (see audioscript_runtime.py for the current entry point)
audioscript_runtime.py	Primary interactive AudioScript shell

Datasets and training metadata live under audio/ai/datasets/ (not a top-level datasets/ folder).

---

AudioScript

AudioScript (AS) is AudioMIX’s musical programming language — human-readable, whitespace-sensitive, and built for emotional, mood-aware control of audio and lighting.

Live vs IR

Branch	Use case
Live	Written on stage; forgiving, expressive syntax (audioscript/live/)
IR	Machine-generated, precise pipelines; compiler output (audioscript/ir/)

Example

mood.set("uplifted")
glow("lilac")
play("intro.wav")
pulse("yellow", bpm=120)

Learn more

• Language specification
• AudioScript directory guide
• Example scripts: audioscript/scripts/

---

Juniper2.0

Juniper2.0 is the AI layer built for AudioMIX — not a generic chatbot, but a collaborator that understands AudioScript, mood mappings, LED profiles, and your session context.

• AudioScript — Generate and debug Live and IR scripts from natural language
• Production — EQ suggestions, track flow, and emotional arc analysis
• Lighting — Map inferred mood to color and pattern profiles
• Live performance — Audience-aware transitions and guidance (roadmap)

Design principle: amplify creative intent; every suggestion is traceable and overridable.

"Juniper doesn't replace the artist — she collaborates with them."

Integration into the CLI and AudioMIX-Electron is in active development.

---

External services

Spotify

• Uses the official Spotify Web API with OAuth — no scraping
• Feature extraction via permitted endpoints per Spotify Developer Terms
• Credentials via .env (see .gitignore; never commit secrets)

Bandcamp

• Embed-based integration for ethical, artist-facing playback links

---

AudioMIX-Electron

Companion desktop UI for the core engine — Electron shell with DAW-style panels (mixer, EQ, LED zones) and a secure preload bridge to the Python/C++ backend.

Layer	Repository
Core engine	audiomix (this repo)
Desktop UI	audiomix-electron

git clone https://github.com/alexisvassquez/audiomix-electron.git
cd audiomix-electron
npm install
npm run dev

Status: First Breath milestone (v0.1-dev). FastAPI bridge to the backend is planned.

---

Development

Python stack

Key dependencies (full list in requirements.txt):

• Audio: librosa, soundfile, sounddevice, PyAudio, mido, python-rtmidi
• ML: scikit-learn, torch, transformers
• Integrations: spotipy
• UX: rich

C++ targets

After cmake --build build:

Target	Role
audiomix	Main DSP binary (used by DSPBridge)
test_sine_dsp	Sanity check tone generator
test_clipper	Clipper module test

Contributing

1. Read CHANGELOG.md for recent architectural decisions
2. Follow the Ethical AI manifesto
3. Prefer extending performance_engine/modules/ and emitting events on performance_engine.event_bus.bus

---

Ethical AI

AudioMIX is developed with an artist-first, emotionally-aware, open-source philosophy.

See ETHICAL_AI_MANIFESTO.md for principles and creative mission.

---

License

GNU General Public License v3 — free for independent artists and developers.

---

© 2026 Alexis M. Vasquez · AMV Digital Studios