Cyclic multichannel Wiener filter (cMWF) for acoustic beamforming

Introduction

This code accompanies the paper "Cyclic multichannel Wiener filter for acoustic beamforming", accepted in WASPAA 2025 (arXiv: https://arxiv.org/abs/2507.10159).

Check also our follow-up work, where almost-periodic noises are suppressed using a cyclic MVDR: https://github.com/Screeen/cmvdr.

The paper proposes a new beamforming method leveraging the cyclostationarity of voiced speech signals.
By exploiting correlations across microphones and frequency components, the cyclic multichannel Wiener filter (cMWF) improves noise reduction in low-SNR scenarios.

---

Installation and first run

Prerequisites

• Python 3.9 (for compatibility with librosa)
• Tested on macOS 14.4.1 but should run on most Linux systems.

Setup

1. Clone the repository:

   git clone git@github.com:Screeen/cyclicMWF.git
   cd cyclicMWF

2. Create and activate a Python virtual environment:

   python3.9 -m venv .venv
   source .venv/bin/activate

3. Install required packages:

   pip install --upgrade pip
   uv sync  # requires the `uv` package for managing virtual environments
   # (or if you prefer:)
   # pip install -r requirements.txt

fast_nls_src compilation

We use the fastF0Nls library for pitch estimation, which requires compilation. See https://github.com/jkjaer/fastF0Nls for instructions.

---

Running experiments

Configurations

Experiment parameters are controlled via YAML files in the configs/ folder. Edit default.yaml and cmwf.yaml to set your desired parameters such as:

• data_type (choose from 'synthetic', 'instruments', or 'speech')
• num_montecarlo_simulations
• ...

Example runs

Run synthetic data experiments:

python main.py

Notes

• Real data experiments require downloading room impulse responses, clean speech, and instrument samples:
• RIRs: https://www.iks.rwth-aachen.de/en/research/tools-downloads/databases/multi-channel-impulse-response-database/
• Instrument samples: https://theremin.music.uiowa.edu/MIS.html.
• Speech: Speech Intelligibility CD by Neil Thompson Shade

Troubleshooting

If you get

ImportError: cannot import name 'kaiser' from 'scipy.signal' (.../.venv/lib/python3.X/site-packages/scipy/signal/__init__.py). Did you mean: 'kaiserord'?

Open the file

nano .../SVD-direct/env/lib/python3.X/site-packages/pysepm/util.py

and replace

from scipy.signal import firls,kaiser,upfirdn

with

from scipy.signal import firls,upfirdn
from scipy.signal.windows import kaiser

---

Note

Feedback and questions welcome: G.Bologni@tudelft.nl. Enjoy experimenting with cMWF!