Deep Noise Suppression (DNS) Challenge - Interspeech 2020

This repository contains the datasets and scripts required for the DNS challenge. For more details about the challenge, please visit https://dns-challenge.azurewebsites.net/ and refer to our paper.

Repo details:

• The datasets directory contains the clean speech and noise clips.
• The NSNet-baseline directory contains the inference scripts and the ONNX model for the baseline Speech Enhancer called Noise Suppression Net (NSNet)
• noisyspeech_synthesizer_singleprocess.py - is used to synthesize noisy-clean speech pairs for training purposes.
• noisyspeech_synthesizer.cfg - is the configuration file used to synthesize the data. Users are required to accurately specify different parameters.
• audiolib.py - contains modules required to synthesize datasets
• utils.py - contains some utility functions required to synthesize the data
• unit_tests_synthesizer.py - contains the unit tests to ensure sanity of the data

Prerequisites

• Python 3.0 and above
• Soundfile (pip install pysoundfile), librosa

Usage:

1. Install librosa

pip install librosa

2. Install Git Large File Storage for faster download of the datasets.

git lfs install
git lfs track "*.wav"
git add .gitattributes

3. Clone the repository.

git clone https://github.com/microsoft/DNS-Challenge DNS-Challenge

4. Edit noisyspeech_synthesizer.cfg to include the paths to clean speech and noise directories. Also, specify the paths to the destination directories and store logs.
5. Create dataset

python noisyspeech_synthesizer_multiprocessing.py

Citation:

For the datasets and the DNS challenge:

@article{reddy2020interspeech,
  title={The INTERSPEECH 2020 Deep Noise Suppression Challenge: Datasets, Subjective Testing Framework, and Challenge Results},
  author={Reddy, Chandan KA and Gopal, Vishak and Cutler, R