AudioDev

Audio DSP development ecosystem for algorithm development, testing, and deployment.

Active Projects

audio-graph-python

Hybrid Faust/C++ audio processing graph system with Python control.

• Flat channel model (all signals mono, multichannel = N parallel signals)
• Uniform interface for Faust and C++ processors
• Named processor access, capture mode for debugging, chunked processing
• Graph specification language for cascade definitions

dsp_library

Framework-agnostic C++ and Faust DSP implementations shared across platforms.

• C++ Processors: example_processor, split_chorus
• Faust Modules: gain, split_gain, dual_tap_delay (9th order Lagrange interpolation)
• Single source of truth: write DSP once, wrap for Python and Max

max_externals

Max/MSP external wrappers for DSP processors.

• dual_tap_delay~: Faust-based dual tap delay
• split_chorus~: C++ split chorus effect
• Thin wrappers that include DSP from dsp_library

Reference Projects

Embedded DSP DevSystem

Conceptual framework for DSP algorithm development lifecycle (Python prototyping to embedded C++). Contains design documents and architectural decisions.

DaisyExamples

Examples for Electro-Smith Daisy embedded audio platform. 80+ DSP algorithm examples.

SciPy

Python-based signal processing algorithm development and analysis.

PitchShifter

Early-stage bass guitar pitch shifter concept.

Architecture

AudioDev/
├── dsp_library/           # Core DSP implementations (C++ and Faust)
│   ├── cpp/               # Framework-agnostic C++ DSP
│   └── faust/             # Faust DSP modules
├── audio-graph-python/    # Python graph system + pybind11 bindings
│   ├── python/            # Graph system
│   ├── bindings/          # pybind11 wrappers for dsp_library
│   └── examples/          # Working examples
├── max_externals/         # Max/MSP external wrappers
│   ├── dual_tap_delay_tilde/
│   └── split_chorus_tilde/
└── ...                    # Reference projects

Design Principles

1. Single source of truth: DSP code lives in dsp_library/, wrapped for each platform
2. Flat channel model: All signals are mono, multichannel is N parallel mono signals
3. Uniform interface: All processors expose init/process/set_param/get_param
4. Framework-agnostic: Core DSP has no dependencies on Max, Python, or any framework

Getting Started

See audio-graph-python/QUICKSTART.md for a 5-minute introduction.

Recent Changes

See audio-graph-python/CHANGELOG.md for recent updates.

Latest (2026-01-15):

• Split chorus C++ module with Python bindings
• Graph specification language for cascade definitions
• File reorganization with dsp_library as central DSP source