Spoofing detection under noisy conditions: a preliminary investigation and an initial database

View on arXiv ← Back to list

Authors: Xiaohai Tian, Zhizheng Wu, Xiong Xiao, Eng Siong Chng, Haizhou Li

Published: 2016-02-09 12:00:56+00:00

AI Summary

This paper investigates spoofing detection for automatic speaker verification (ASV) under noisy conditions. A new database is created by adding various noises to the ASVspoof 2015 database at different signal-to-noise ratios (SNRs), and experiments show that system performance degrades significantly under noisy conditions, with phase-based features proving more robust than magnitude-based features.

Abstract

Spoofing detection for automatic speaker verification (ASV), which is to discriminate between live speech and attacks, has received increasing attentions recently. However, all the previous studies have been done on the clean data without significant additive noise. To simulate the real-life scenarios, we perform a preliminary investigation of spoofing detection under additive noisy conditions, and also describe an initial database for this task. The noisy database is based on the ASVspoof challenge 2015 database and generated by artificially adding background noises at different signal-to-noise ratios (SNRs). Five different additive noises are included. Our preliminary results show that using the model trained from clean data, the system performance degrades significantly in noisy conditions. Phase-based feature is more noise robust than magnitude-based features. And the systems perform significantly differ under different noise scenarios.

Key findings

Spoofing detection performance degrades significantly in noisy conditions compared to clean data. Phase-based features are more robust to noise than magnitude-based features. Performance varies significantly across different noise types and SNRs.

Approach

The authors added five types of noise to the ASVspoof 2015 database at three SNR levels to create a noisy database. They then evaluated a state-of-the-art spoofing detection system, using features extracted from the noisy data and a classifier trained on clean data, measuring performance using equal error rate (EER).

Datasets

ASVspoof 2015 challenge database, NOISEX-92 database, QUT-NOISE database

Model(s)

Multilayer Perceptron (MLP)

Author countries

Singapore, United Kingdom

← Previous