Deep MOS Predictor for Synthetic Speech Using Cluster-Based Modeling

TL;DR

This paper introduces three cluster-based deep learning models to improve automated speech quality assessment, outperforming previous models like MOSNet by better predicting human judgments in speech synthesis evaluation.

Contribution

It proposes three novel cluster-based models incorporating GQT and Encoding Layers to enhance speech quality and similarity score predictions.

Findings

GQT layer improves prediction accuracy of human assessments.

Encoding Layer enhances utilization of frame-level scores.

Models outperform previous assessment methods on Voice Conversion Challenge data.

Abstract

While deep learning has made impressive progress in speech synthesis and voice conversion, the assessment of the synthesized speech is still carried out by human participants. Several recent papers have proposed deep-learning-based assessment models and shown the potential to automate the speech quality assessment. To improve the previously proposed assessment model, MOSNet, we propose three models using cluster-based modeling methods: using a global quality token (GQT) layer, using an Encoding Layer, and using both of them. We perform experiments using the evaluation results of the Voice Conversion Challenge 2018 to predict the mean opinion score of synthesized speech and similarity score between synthesized speech and reference speech. The results show that the GQT layer helps to predict human assessment better by automatically learning the useful quality tokens for the task and that the Encoding Layer helps to utilize frame-level scores more precisely.