基于梅尔频率倒谱系数的语音清晰度DRT识别

Speech Articulation DRT Recognition Based On Mel Frequency Cepstral Coefficient

语音清晰度在通信终端、设备系统语音识别方面具有重要意义。本文对110dB噪声干扰下采集到的语音信号进行谱减法降噪,双门限端点检测提取发音字段,然后提取梅尔频率倒谱系数(MFCC),再将其进行差分计算,得到一阶和二阶分量,结合短时能量作为语音信号的特征参数,最后通过动态时间归整(DTW)进行相似度识别。实验表明,本文算法对汉语清晰度诊断押韵测试(DRT)字表的测试结果高达92.90%,有良好的识别率。

Speech articulation plays an important role in speech recognition of communication terminals and equipment systems.In this paper,under the interference of 110 dB noise,the collected speech signal is denoised by spectral subtraction,the pronunciation field is extracted by double-threshold endpoint detection,and then the Mel Frequency Cepstral Coefficients(MFCC)is extracted,and the difference calculation is carried out to obtain the first-order and second-order components,and the short-time energy is used as the characteristic parameter of the speech signal.Finally,Dynamic Time Warping(DTW)is used for similarity recognition.The experimental results show that the algorithm has a high recognition rate of 92.90%for Chinese articulation Diagnostic Rhyme Test(DRT).