基于声学参数测量的语音清晰度预测系统

Speech Intelligibility Prediction System Based on Acoustics Measurements

以信噪比、混响时间和系统截止频率为特征值,利用粒子群算法优化的支持向量机在MATLAB平台下设计了一套语音清晰度预测系统,实现了信噪比、混响时间和系统幅频曲线集成测量与语音清晰度的预测。

In this paper, signal to noise ratio, reverberation time and system cutoff frequency are used as eigenvalues. And the speech intelligibility prediction system is designed based on support vector machines which are optimized by particle swarm optimization under the MATLAB platform. The system realizes the integrated measurement of signal-to-noise ratio, reverberation time, system amplitude-frequency curve and prediction of speech intelligibility.