Skip to content

Latest commit

 

History

History
66 lines (48 loc) · 1.69 KB

README.md

File metadata and controls

66 lines (48 loc) · 1.69 KB
Raw

PHASEN

Unofficial PyTorch implementation of MSRA's: PHASEN: A Phase-and-Harmonics-Aware Speech Enhancement Network.

My resutls on real-world test

Noisy enh

Maybe there is something different with the paper, but it worked not bad.

how to use it?

  1. install dependency:
pip install -r requirements.txt
  1. download datasets

if you don't have WSJ0, you can follow this use aishell-1 by following this se-cldnn-torch

Attetion

There is something different from se-cldnn-torch: the two list for train (tr.lst, cv.lst ...) need duration information, but se-cldnn-torch dose not need it (because the two dataset.py are different).

So, in this repo, train and cross-validation list nead to be like this

/path/noisy1.wav /path/ref1.wav 3.0233
/path/noisy2.wav /path/ref2.wav 2.3213
/path/noisy2.wav /path/ref2.wav 8.8127
...

To add duration information, you can use tools/add_duration.py like:

python tools/add_duration.py data/tr_wsj0.lst

As for inference stage (decode stage, eval stage), the list only need the path of noisy path:

/path/noisy1.wav
/path/noisy2.wav
/path/noisy2.wav
...
  1. run. before you run it, please set the correct params in ./run_phasen.sh
bash run_phasen.sh

Reference:

funcwj's voice-filter

wangkenpu's Conv-Tasnet

pseeth's torch-stft