Deep MOS Predictor for Synthetic Speech Using Cluster-Based Modeling

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.

arXiv comment5 pages, 1 figure, accepted to Interspeech 2020

References in corpus (5)

• WaveNet: A Generative Model for Raw Audio
• Style Tokens: Unsupervised Style Modeling, Control and Transfer in End-to-End Speech Synthesis
• MOSNet: Deep Learning based Objective Assessment for Voice Conversion
• AutoMOS: Learning a non-intrusive assessor of naturalness-of-speech
• Spatial Pyramid Encoding with Convex Length Normalization for Text-Independent Speaker Verification

Cited by in corpus (4)

• Learning to Maximize Speech Quality Directly Using MOS Prediction for Neural Text-to-Speech
• MBNet: MOS Prediction for Synthesized Speech with Mean-Bias Network
• Neural MOS Prediction for Synthesized Speech Using Multi-Task Learning With Spoofing Detection and Spoofing Type Classification
• Wideband Audio Waveform Evaluation Networks: Efficient, Accurate Estimation of Speech Qualities