Pytorch code for following paper:
- Title : Extended U-Net for Speaker Verification in Noisy Environments (Oral presentation at Interspeech2022)
- Autor : Ju-ho Kim, Jungwoo Heo, Hye-jin Shim, and Ha-Jin Yu
Background noise is a well-known factor that deteriorates the accuracy and reliability of speaker verification (SV) systems by blurring speech intelligibility.
Various studies have used separate pretrained enhancement models as the front-end module of the SV system in noisy environments, and these methods effectively remove noises.
However, the denoising process of independent enhancement models not tailored to the SV task can also distort the speaker information included in utterances.
We argue that the enhancement network and speaker embedding extractor should be fully jointly trained for SV tasks under noisy conditions to alleviate this issue.
Therefore, we proposed a U-Net-based integrated framework that simultaneously optimizes speaker identification and feature enhancement losses.
Moreover, we analyzed the structural limitations of using U-Net directly for noise SV tasks and further proposed Extended U-Net to reduce these drawbacks.
We evaluated the models on the noise-synthesized VoxCeleb1 test set and VOiCES development set recorded in various noisy scenarios.
The experimental results demonstrate that the U-Net-based fully joint training framework is more effective than the baseline, and the extended U-Net exhibited state-of-the-art performance versus the recently proposed compensation systems.
- We used 'nvcr.io/nvidia/pytorch:21.04-py3' image of Nvidia GPU Cloud for conducting our experiments.
- We used three Titan RTX GPUs for training.
- Python 3.6.9
- Pytorch 1.8.1
- Torchaudio 0.8.1
We used VoxCeleb1 dataset for training and test.
For noise synthesis, we used the MUSAN corpus.
Go into run directory
./train.sh
@article{kim2022extended,
title={Extended U-Net for Speaker Verification in Noisy Environments},
author={Kim, Ju-ho and Heo, Jungwoo and Shim, Hye-jin and Yu, Ha-Jin},
journal={arXiv preprint arXiv:2206.13044},
year={2022}
}
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent