Skip to content

CLAUDE.md

Latest commit

 

History

History
72 lines (54 loc) · 2.59 KB

CLAUDE.md

File metadata and controls

72 lines (54 loc) · 2.59 KB
Raw

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Build and Development Commands

# Install in development mode
pip install -e .

# Install with all optional dependencies
pip install -e ".[full]"

# Install specific extras
pip install -e ".[synth]"   # Synthesis modules
pip install -e ".[midi]"    # MIDI processing (mido)
pip install -e ".[ml]"      # Machine learning (scikit-learn, umap)
pip install -e ".[viz]"     # Visualization (matplotlib)
pip install -e ".[audio]"   # Extended audio (librosa, pydub)

# Run tests
python -m pytest tests/

# Run a single test file
python tests/t_subtractive_synth.py

Architecture

Package structure: audio_dsp with five main submodules that use lazy imports to avoid requiring all optional dependencies:

  • synth - Synthesizers (SubtractiveSynth, DX7FMSynth, DrumSynth, chip tones, Karplus-Strong)
  • effects - Audio effects with unified API: effect(signal, sample_rate, **params) → numpy.ndarray
  • sequencer - Algorithmic composition (raga generator, tree/matrix composers, Game of Life sequencers)
  • midi - MIDI utilities requiring mido (polyrhythmic generation, looping, alternate tunings)
  • utils - Audio I/O, scales, noise algorithms, spectral analysis

Core dependencies: numpy, scipy, soundfile (always available)

Optional dependencies: librosa, mido, matplotlib, scikit-learn, opencv-python, simpleaudio

Key Patterns

Effects unified API

All effects follow: effect_name(signal, sample_rate, **params) -> numpy.ndarray

from audio_dsp.effects import filter, delay, reverb
audio = filter(audio, sr, cutoff=1000, filter_type="lowpass")
audio = delay(audio, sr, delay_time=0.25, feedback=0.5)

Backward compatibility aliases

Effects module exports both new names (delay, compress, autotune) and legacy aliases (SuperDelay, SuperCleanCompressor, autotune_effect).

Synthesizer pattern

Synths are classes with configurable properties and a synthesize(freq, duration, **envelope) method:

synth = SubtractiveSynth(sample_rate=44100)
synth.osc_wave = "saw"
synth.filter_cutoff = 800
audio = synth.synthesize(freq=220, duration=2.0)

Audio I/O

load_audio() returns (sample_rate, audio_array) - note the scipy convention order.

from audio_dsp.utils import load_audio
sr, audio = load_audio("file.wav", mono=True)

Try/except import pattern

Submodules wrap optional dependencies in try/except blocks. When adding features requiring optional deps, follow this pattern in __init__.py files.