mindslab-ai/voicefilter

Unofficial PyTorch implementation of Google AI’s VoiceFilter system


repo name	mindslab-ai/voicefilter
repo link	https://github.com/mindslab-ai/voicefilter
homepage	http://swpark.me/voicefilter
language	Python
size (curr.)	1195 kB
stars (curr.)	445
created	2019-03-22
license

VoiceFilter

Unofficial PyTorch implementation of Google AI’s: VoiceFilter: Targeted Voice Separation by Speaker-Conditioned Spectrogram Masking.

Result

Training took about 20 hours on AWS p3.2xlarge(NVIDIA V100).

Audio Sample

Listen to audio sample at webpage: http://swpark.me/voicefilter/

Metric

Median SDR	Paper	Ours
before VoiceFilter	2.5	1.9
after VoiceFilter	12.6	10.2

SDR converged at 10, which is slightly lower than paper’s.

Dependencies

Python and packages

This code was tested on Python 3.6 with PyTorch 1.0.1. Other packages can be installed by:
```
pip install -r requirements.txt
```
Miscellaneous

ffmpeg-normalize is used for resampling and normalizing wav files. See README.md of ffmpeg-normalize for installation.

Prepare Dataset

Download LibriSpeech dataset

To replicate VoiceFilter paper, get LibriSpeech dataset at http://www.openslr.org/12/. train-clear-100.tar.gz(6.3G) contains speech of 252 speakers, and train-clear-360.tar.gz(23G) contains 922 speakers. You may use either, but the more speakers you have in dataset, the more better VoiceFilter will be.

Resample & Normalize wav files

First, unzip tar.gz file to desired folder:

tar -xvzf train-clear-360.tar.gz

Next, copy utils/normalize-resample.sh to root directory of unzipped data folder. Then:

vim normalize-resample.sh # set "N" as your CPU core number.
chmod a+x normalize-resample.sh
./normalize-resample.sh # this may take long

Edit config.yaml

cd config
cp default.yaml config.yaml
vim config.yaml

Preprocess wav files

In order to boost training speed, perform STFT for each files before training by:
```
python generator.py -c [config yaml] -d [data directory] -o [output directory] -p [processes to run]
```
This will create 100,000(train) + 1000(test) data. (About 160G)

Train VoiceFilter

Get pretrained model for speaker recognition system

VoiceFilter utilizes speaker recognition system (d-vector embeddings). Here, we provide pretrained model for obtaining d-vector embeddings.

This model was trained with VoxCeleb2 dataset, where utterances are randomly fit to time length [70, 90] frames. Tests are done with window 80 / hop 40 and have shown equal error rate about 1%. Data used for test were selected from first 8 speakers of VoxCeleb1 test dataset, where 10 utterances per each speakers are randomly selected.

Update: Evaluation on VoxCeleb1 selected pair showed 7.4% EER.

The model can be downloaded at this GDrive link.
Run

After specifying train_dir, test_dir at config.yaml, run:
```
python trainer.py -c [config yaml] -e [path of embedder pt file] -m [name]
```
This will create chkpt/name and logs/name at base directory(-b option, . in default)
View tensorboardX
```
tensorboard --logdir ./logs
```

Resuming from checkpoint

python trainer.py -c [config yaml] --checkpoint_path [chkpt/name/chkpt_{step}.pt] -e [path of embedder pt file] -m name

Evaluate

python inference.py -c [config yaml] -e [path of embedder pt file] --checkpoint_path [path of chkpt pt file] -m [path of mixed wav file] -r [path of reference wav file] -o [output directory]