WaoN Python Bindings

Python bindings for WaoN - Wave-to-Notes transcriber.

Installation

From source

pip install maturin
maturin develop --release

From wheel

pip install waon

Usage

Basic transcription

import waon
import numpy as np

# Transcribe an audio file
midi_bytes = waon.transcribe_file("audio.wav")
with open("output.mid", "wb") as f:
    f.write(midi_bytes)

# Transcribe audio data
audio_data = np.array([...], dtype=np.float32)
midi_bytes = waon.transcribe(audio_data, sample_rate=44100)

Advanced usage with options

import waon

# Create custom options
options = waon.WaonOptions()
options.fft_size = 4096
options.hop_size = 2048
options.window_type = "hamming"
options.note_bottom = 36  # C2
options.note_top = 96     # C7
options.velocity_threshold = 0.005
options.use_phase_vocoder = True

# Create context with options
context = waon.WaonContext(
    fft_size=4096,
    hop_size=2048,
    window_type="hamming"
)

# Transcribe
midi_bytes = context.transcribe_file("audio.wav")

# Analyze a single frame
audio_frame = np.array([...], dtype=np.float32)
note_intensities = context.analyze(audio_frame)  # Returns 128-element array

API Reference

Functions

• transcribe(audio, sample_rate=44100, options=None) - Transcribe audio array to MIDI
• transcribe_file(path, options=None) - Transcribe audio file to MIDI
• get_version() - Get library version

Classes

• WaonContext - Transcription context for stateful processing
• WaonOptions - Configuration options for transcription

Window Types

• hanning (default)
• hamming
• blackman
• nuttall
• blackman2, blackman3, blackman4
• exponentialsine
• parzen
• welch
• steeper30, steeper60

License

GPL-2.0-or-later