Adaptif

Rust crate for DSP Adaptive Filters with Python bindings.

Overview

This project contains Rust implementations of adaptive filtering algorithms, with bindings for Python. Currently, the following algorithms are supported:

We're also working on real-time processing and no_std functionality for use in embedded systems, which are planned for the next major release.

Installation

The projects currently supports Rust stable (no MSRV) and Python 3.11 - 3.15.

Rust crate

Via crates.io:

cargo add adaptif

Python bindings

Via PyPI:

# With uv (preferred).
uv add adaptif

# With pip.
pip install adaptif

Usage

This section shows basic examples for a least mean squares (LMS) filter in Rust and Python.

Here:

• input_signal is the desired/noisy signal. Commonly referred to as d(n) in mathematical notation.

• noise_ref is the reference noise signal. Commonly referred to as x(n) in mathematical notation.

More examples can be found in examples/.

Rust

use adaptif::filters::{LmsFilter, LmsOptions, ProcessingMode};
use adaptif::types::{InputSignal, NoiseReference};

// Sample inputs
let input_signal = InputSignal::new(vec![1.0, -2.5, 3.0])?;
let noise_ref = NoiseReference::new(vec![2.0, -1.2, -3.8])?;

// Initialize the filter
let mut lms = LmsFilter::new(
    LmsOptions {
        step_size: 0.1,  // How fast the filter adapts
        window_size: 16, // How many samples the filter processes at a time
    },
    ProcessingMode::Sample,
)?;

// Adapt the filter to the inputs and get the iteratively cleaned signal.
let cleaned_signal = lms.adapt(&input_signal, &noise_ref)?;

Ok(())

Python

from adaptif import LMSFilter

# setting up filter
lms = LMSFilter(step_size=0.1, window_size=16)

# Assuming signals are already present and named accordingly
cleaned_signal = lms.adapt(input_signal=noisy_signal, noise_ref=noise)

Contributing

See CONTRIBUTING.md for setup and contribution guidelines.

In short,

• Open an issue for a discussion. We will likely handle it.
• Undisclosed AI PRs will be closed and no further PRs from said user will be considered.

Citation

If you found any of this helpful, feel free to cite it, or just send us an email.

@misc{adaptif,
  authors = {Benjamin Glidden, Elias Naske},
  title = {Adaptif: Rust crate for DSP adaptive filters with Python bindings},
  year = {2026},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/bglid/adaptif}}
}