Detection of Doctored Speech: Towards an End-to-End Parametric Learn-able Filter Approach

View on arXiv ← Back to list

Authors: Rohit Arora

Published: 2022-06-27 06:28:46+00:00

AI Summary

This research proposes end-to-end deep learning models (WSTnet and CWTnet) for detecting doctored speech, using Wavelet Scattering and Continuous Wavelet Transforms, respectively, instead of the SincNet baseline's sinc layer. A further improved model, WDnet, replaces the CWT layer with a Wavelet Deconvolution layer to optimize scale parameters, yielding substantial performance improvements over both the baseline and traditional methods on the ASVspoof 2019 dataset.

Abstract

The Automatic Speaker Verification systems have potential in biometrics applications for logical control access and authentication. A lot of things happen to be at stake if the ASV system is compromised. The preliminary work presents a comparative analysis of the wavelet and MFCC-based state-of-the-art spoof detection techniques developed in these papers, respectively (Novoselov et al., 2016) (Alam et al., 2016a). The results on ASVspoof 2015 justify our inclination towards wavelet-based features instead of MFCC features. The experiments on the ASVspoof 2019 database show the lack of credibility of the traditional handcrafted features and give us more reason to progress towards using end-to-end deep neural networks and more recent techniques. We use Sincnet architecture as our baseline. We get E2E deep learning models, which we call WSTnet and CWTnet, respectively, by replacing the Sinc layer with the Wavelet Scattering and Continuous wavelet transform layers. The fusion model achieved 62% and 17% relative improvement over traditional handcrafted models and our Sincnet baseline when evaluated on the modern spoofing attacks in ASVspoof 2019. The final scale distribution and the number of scales used in CWTnet are far from optimal for the task at hand. So to solve this problem, we replaced the CWT layer with a Wavelet Deconvolution(WD) (Khan and Yener, 2018) layer in our CWTnet architecture. This layer calculates the Discrete-Continuous Wavelet Transform similar to the CWTnet but also optimizes the scale parameter using back-propagation. The WDnet model achieved 26% and 7% relative improvement over CWTnet and Sincnet models respectively when evaluated over ASVspoof 2019 dataset. This shows that more generalized features are extracted as compared to the features extracted by CWTnet as only the most important and relevant frequency regions are focused upon.

Key findings

The proposed wavelet-based end-to-end models, particularly WDnet, achieved significant relative improvements in performance (up to 62% and 26%) over traditional handcrafted models and the SincNet baseline on the ASVspoof 2019 dataset. Fusion of multiple models also improved results, suggesting that combining different spectral representations enhances spoof detection.

Approach

The authors replace the Sinc layer in the SincNet architecture with Wavelet Scattering and Continuous Wavelet Transform layers to create WSTnet and CWTnet respectively. To further optimize feature extraction, they introduce WDnet, substituting the CWT layer with a Wavelet Deconvolution layer for learning-based scale parameter optimization.

Datasets

ASVspoof 2015 and ASVspoof 2019 datasets

Model(s)

SincNet, WSTnet, CWTnet, WDnet. Gaussian Mixture Models (GMMs) were used as a classifier in some experiments with handcrafted features.

Author countries

India

← Previous