PyTorch Implementation of the model described in UTTERANCE-LEVEL AGGREGATION FOR SPEAKER RECOGNITION IN THE WILD: https://arxiv.org/pdf/1902.10107.pdf

Create the conda environment:

conda env create -f environment.yml 
conda activate SV_pytorch

Download the Vox Celeb dataset from here: http://www.robots.ox.ac.uk/~vgg/data/voxceleb/vox1.html

Once you download the dataset specify the proper path instead of PATHTODATASET in config/config.yml:

train_path: './PATHTODATASET/Audio/dev/processed'
train_path_unprocessed: '/PATHTODATASET/Audio/dev/wav' #*/wav/speaker_id/session_id/file.wav
test_meta_path: './dataVoxCeleb1/Metadata/veri_test.txt'
test_path: '/PATHTODATASET/Audio/test/processed'
test_path_unprocessed: '/PATHTODATASET/Audio/test/wav'

To convert the wav files to spectograms run the following utility function:

python data_preprocess.py

Two folders will be created, dev/processed and test/processed, containing .npy files containing numpy ndarrays of speaker utterances and test pair utterances, respectively.

To train the speaker verification model, run:

python train_speech_embedder.py 

with the following config.yaml key set to true:

training: !!bool "true"

for testing, set the key value to:

training: !!bool "false"

The log file and checkpoint save locations are controlled by the following values:

log_file: './speech_id_checkpoint/Stats'
checkpoint_dir: './speech_id_checkpoint'