Real-time measurement for experience quality of underwater acoustic voice communication

Abstract

The real-time measurement of quality of the underwater acoustic voice communications is the important link to safeguard the quality of communication. Using real-time measurements, voice modulation parameters can be adjusted in a timely manner and the adaptive ability of the link can be improved. This paper proposes a kind of objective assessment model of voice quality based on parameter extraction and underwater acoustic channel characteristics. Three feature parameters of voice are extracted in this model. They are MFCC, LPCC and LSD. These three parameters form the spectral distortion evaluation. Due to the high correlation between PESQ (Perceptual Evaluation of Speech Quality) and MOS, this paper adopts PESQ-MOS instead of MOS and distortion evaluation of spectral feature parameters to establish a mapping function. In this way, the complex and time-consuming manual evaluation process can be avoided, and the similarity of evaluation results and subjective feelings can be improved. The mapping relationship of distortion evaluation and receiving voice quality PESQ-MOS (Perceptual Evaluation of researched and Quality-Mean Opinion Score) is used to realize quantization. The paper also introduces the dynamic MFCC spectral distortion evaluation as a regulatory factor. It helps the process to be more adaptable. In general, this paper mainly studies the quality estimation model of FM underwater acoustic voice communication. The performance of the estimation model is verified by means of simulation experiments using a speech sample library. The practical performances of experiments in more complex environments are evaluated. The results of experiments show that the performance of the voice quality evaluation model proposed in this paper is similar to subjective feeling.