pyminidsp 0.6.2

Python bindings to the miniDSP C library

pyminidsp

AI agents: fetch llms-full.txt for the complete API reference and tutorials in a single markdown file.

Documentation

Python bindings to the miniDSP C library.

A comprehensive DSP toolkit providing signal generation, spectral analysis, filtering, effects, and more. All functions accept and return NumPy arrays.

Installation

Pre-built wheels are available for Linux x86-64 and macOS ARM64 (Apple Silicon), with Python 3.9–3.13. Wheels include the compiled C extension (no compiler needed), but FFTW3 must be installed on your system (apt install libfftw3-3 or brew install fftw).

pip install pyminidsp

Building from Source

If you need to build from source (e.g. unsupported platform or development):

Prerequisites:

• FFTW3 development headers
  • Ubuntu/Debian: sudo apt install libfftw3-dev
  • macOS: brew install fftw
• A C compiler (gcc or clang)

# Clone the miniDSP C library
git clone https://github.com/wooters/miniDSP.git

# Create virtual environment
uv venv

# Install pyminidsp (set MINIDSP_SRC to point to the C library)
MINIDSP_SRC=./miniDSP uv sync

For development (includes test dependencies):

MINIDSP_SRC=./miniDSP uv sync --extra dev

Quick Start

import pyminidsp as md

# Generate a 440 Hz sine wave (1 second at 44.1 kHz)
signal = md.sine_wave(44100, amplitude=1.0, freq=440.0, sample_rate=44100.0)

# Compute the magnitude spectrum
mag = md.magnitude_spectrum(signal)

# Compute MFCCs from a frame
coeffs = md.mfcc(signal[:512], sample_rate=44100.0, num_mels=26, num_coeffs=13)

# Apply a low-pass biquad filter
lpf = md.BiquadFilter(md.LPF, freq=1000.0, sample_rate=44100.0)
filtered = lpf.process_array(signal)

# Clean up FFT caches when done
md.shutdown()

API Overview

Signal Generators

Function	Description
sine_wave(n, amplitude, freq, sample_rate)	Pure sine tone
white_noise(n, amplitude, seed)	Gaussian white noise
impulse(n, amplitude, position)	Kronecker delta
chirp_linear(n, amplitude, f_start, f_end, sample_rate)	Linear frequency sweep
chirp_log(n, amplitude, f_start, f_end, sample_rate)	Logarithmic frequency sweep
square_wave(n, amplitude, freq, sample_rate)	Square wave
sawtooth_wave(n, amplitude, freq, sample_rate)	Sawtooth wave
shepard_tone(n, amplitude, base_freq, sample_rate, rate, num_octaves)	Shepard tone illusion

Spectral Analysis

Function	Description
magnitude_spectrum(signal)	One-sided magnitude spectrum via FFT
power_spectral_density(signal)	Periodogram PSD estimate
phase_spectrum(signal)	Phase spectrum in radians
stft(signal, n, hop)	Short-Time Fourier Transform
stft_num_frames(signal_len, n, hop)	Number of STFT frames
mel_filterbank(n, sample_rate, num_mels)	Mel-spaced triangular filterbank
mel_energies(signal, sample_rate, num_mels)	Mel-band energies
mfcc(signal, sample_rate, num_mels, num_coeffs)	Mel-frequency cepstral coefficients

Window Functions

Function	Description
hann_window(n)	Hann window
hamming_window(n)	Hamming window
blackman_window(n)	Blackman window
rect_window(n)	Rectangular window
kaiser_window(n, beta)	Kaiser window (configurable shape)

Math Utilities

Function	Description
bessel_i0(x)	Zeroth-order modified Bessel function I₀
sinc(x)	Normalized sinc function

Signal Measurement

Function	Description
dot(a, b)	Dot product of two vectors
entropy(signal)	Normalised entropy of a distribution
energy(signal)	Signal energy
power(signal) / power_db(signal)	Signal power (linear / dB)
rms(signal)	Root mean square

Signal Analysis

Function	Description
zero_crossing_rate(signal)	Zero-crossing rate
autocorrelation(signal, max_lag)	Normalised autocorrelation
peak_detect(signal, threshold, min_distance)	Local maxima detection
f0_autocorrelation(signal, sample_rate, min_freq, max_freq)	F0 via autocorrelation
f0_fft(signal, sample_rate, min_freq, max_freq)	F0 via FFT peak picking
mix(signal_a, signal_b, weight_a, weight_b)	Weighted sum of two signals

Signal Scaling

Function	Description
scale(value, old_min, old_max, new_min, new_max)	Map a value from one range to another
scale_vec(signal, old_min, old_max, new_min, new_max)	Map vector elements between ranges
fit_within_range(signal, new_min, new_max)	Fit signal values within a range
adjust_dblevel(signal, target_db)	Automatic gain control

Effects & Filters

Function	Description
delay_echo(signal, delay_samples, feedback, dry, wet)	Echo effect
tremolo(signal, rate_hz, depth, sample_rate)	Amplitude modulation
comb_reverb(signal, delay_samples, feedback, dry, wet)	Comb-filter reverb
convolution_time(signal, kernel)	Time-domain convolution
convolution_fft_ola(signal, kernel)	FFT overlap-add convolution
convolution_num_samples(signal_len, kernel_len)	Output length of linear convolution
moving_average(signal, window_len)	Moving average filter
fir_filter(signal, coeffs)	FIR filter
design_lowpass_fir(num_taps, cutoff, sr, beta)	Design Kaiser-windowed lowpass FIR
lowpass_brickwall(signal, cutoff_hz, sr)	FFT-based ideal lowpass filter
BiquadFilter(type, freq, sample_rate)	IIR biquad filter (LPF/HPF/BPF/etc.)

Resampling

Function	Description
resample(signal, in_rate, out_rate)	Polyphase sinc resampler
resample_output_len(input_len, in_rate, out_rate)	Compute resampled output length

DTMF

Function	Description
dtmf_generate(digits, sample_rate, tone_ms, pause_ms)	Generate DTMF tones
dtmf_detect(signal, sample_rate)	Detect DTMF digits
dtmf_signal_length(num_digits, sample_rate, tone_ms, pause_ms)	Calculate samples needed for DTMF

Delay Estimation (GCC-PHAT)

Function	Description
get_delay(sig_a, sig_b, margin, weighting)	Delay between two signals
get_multiple_delays(signals, margin, weighting)	Delays from reference to M-1 signals
gcc(sig_a, sig_b, weighting)	Full cross-correlation

Audio Steganography

Function	Description
steg_encode(host, message, sample_rate, method)	Hide text in audio
steg_decode(stego, sample_rate, method)	Recover hidden text
steg_encode_bytes(host, data, sample_rate, method)	Hide binary data
steg_decode_bytes(stego, sample_rate, method)	Recover binary data
steg_detect(signal, sample_rate)	Detect steganography method
steg_capacity(host, sample_rate, method)	Maximum message length

Spectrogram Art

Function	Description
spectrogram_text(text, freq_lo, freq_hi, duration, sample_rate)	Text visible in spectrogram

Constants

Constant	Description
LPF, HPF, BPF, NOTCH, PEQ, LSH, HSH	Biquad filter types
STEG_LSB, STEG_FREQ_BAND, STEG_SPECTEXT	Steganography methods
STEG_TYPE_TEXT, STEG_TYPE_BINARY	Steganography payload types
GCC_SIMP, GCC_PHAT	GCC weighting modes

Running Tests

MINIDSP_SRC=./miniDSP uv run pytest tests/ -v

License

MIT