Using Multi-Resolution Feature Maps with Convolutional Neural Networks for Anti-Spoofing in ASV

TL;DR

This paper introduces a multi-resolution spectrogram approach using CNNs for anti-spoofing in speaker verification, enhancing discriminative features while maintaining low computational costs.

Contribution

It proposes stacking spectrograms with different window lengths as multi-resolution inputs for CNNs, improving anti-spoofing performance over single-resolution methods.

Findings

Outperforms score fusion methods on ASVspoof 2019 dataset

Improves frequency and time resolution in feature maps

Maintains low computational costs

Abstract

This paper presents a simple but effective method that uses multi-resolution feature maps with convolutional neural networks (CNNs) for anti-spoofing in automatic speaker verification (ASV). The central idea is to alleviate the problem that the feature maps commonly used in anti-spoofing networks are insufficient for building discriminative representations of audio segments, as they are often extracted by a single-length sliding window. Resulting trade-offs between time and frequency resolutions restrict the information in single spectrograms. The proposed method improves both frequency resolution and time resolution by stacking multiple spectrograms that are extracted using different window lengths. These are fed into a convolutional neural network in the form of multiple channels, making it possible to extract more information from input signals while only marginally increasing computational costs. The efficiency of the proposed method has been conformed on the ASVspoof 2019 database. We show that the use of the proposed multiresolution inputs consistently outperforms that of score fusion across different CNN architectures. Moreover, computational cost remains small.