Full-Reference Speech Quality Estimation with Attentional Siamese Neural Networks

TL;DR

This paper introduces a deep learning model using attentional Siamese neural networks for full-reference speech quality prediction, effectively aligning and assessing degraded speech signals against their clean references.

Contribution

It presents a novel neural network architecture that incorporates attention and Siamese structures for improved speech quality estimation with reference signals.

Findings

Effective alignment of speech signals using attention mechanisms.

Improved speech quality prediction accuracy.

Simple solution for time-alignment in VoIP speech signals.

Abstract

In this paper, we present a full-reference speech quality prediction model with a deep learning approach. The model determines a feature representation of the reference and the degraded signal through a siamese recurrent convolutional network that shares the weights for both signals as input. The resulting features are then used to align the signals with an attention mechanism and are finally combined to estimate the overall speech quality. The proposed network architecture represents a simple solution for the time-alignment problem that occurs for speech signals transmitted through Voice-Over-IP networks and shows how the clean reference signal can be incorporated into speech quality models that are based on end-to-end trained neural networks.