The official PyTorch implementation of "Inter-SubNet: Speech Enhancement with Subband Interaction", accepted by ICASSP 2023.
📜[Full Paper] ▶[Demo] 💿[Checkpoint]
Linux or macOS
python>=3.6
Anaconda or Miniconda
NVIDIA GPU + CUDA CuDNN (CPU can also be supported)
Install Anaconda or Miniconda, and then install conda and pip packages:
# Create conda environment
conda create --name speech_enhance python=3.8
conda activate speech_enhance
# Install conda packages
# Check python=3.8, cudatoolkit=10.2, pytorch=1.7.1, torchaudio=0.7
conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch
conda install tensorboard joblib matplotlib
# Install pip packages
# Check librosa=0.8
pip install Cython
pip install librosa pesq pypesq pystoi tqdm toml colorful mir_eval torch_complex
# (Optional) If you want to load "mp3" format audio in your dataset
conda install -c conda-forge ffmpeg
Clone the repository:
git clone https://github.com/RookieJunChen/Inter-SubNet.git
cd Inter-SubNet
Download the pre-trained checkpoint, and input commands:
source activate speech_enhance
python -m speech_enhance.tools.inference \
-C config/inference.toml \
-M $MODEL_DIR \
-I $INPUT_DIR \
-O $OUTPUT_DIR
git clone https://github.com/RookieJunChen/Inter-SubNet.git
cd Inter-SubNet
Please prepare your data in the data dir as like:
and set the train dir in the script run.sh
.
Then:
source activate speech_enhance
bash run.sh 0 # peprare training list or meta file
Please prepare your test cases dir like: data/test_cases_<name>
, and set the test dir in the script run.sh
.
First, you need to modify the various configurations in config/train.toml
for training.
Then you can run training:
source activate speech_enhance
bash run.sh 1
After training, you can enhance noisy speech. Before inference, you first need to modify the configuration in config/inference.toml
.
You can also run inference:
source activate speech_enhance
bash run.sh 2
Or you can just use inference.sh
:
source activate speech_enhance
bash inference.sh
Calculating objective metrics (SI_SDR, STOI, WB_PESQ, NB_PESQ, etc.) :
bash metrics.sh
For test set without reference, you can obtain subjective scores (DNS_MOS and NISQA, etc) through DNSMOS and NISQA.
If you find our work useful in your research, please consider citing:
@inproceedings{chen2023inter,
title={Inter-Subnet: Speech Enhancement with Subband Interaction},
author={Chen, Jun and Rao, Wei and Wang, Zilin and Lin, Jiuxin and Wu, Zhiyong and Wang, Yannan and Shang, Shidong and Meng, Helen},
booktitle={ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
pages={1--5},
year={2023},
organization={IEEE}
}