基于RASTA-FF2滤波降噪技术的语音识别

语音识别技术可以为要求双手同时作业的操作人员和残疾人提供一种便捷的控制方法.本文提出了一种通过结合FF2(Second-order Frequency Filtering)和RASTA(RelAtive SpecTrAl)技术来增强语音识别鲁棒性的方法,并将这种方法成功应用于机器人化护理床的控制系统中,增强了识别系统在医院、工厂等非稳定噪声环境下语音识别的鲁棒性.通过将HMM/GMM混合模型的传统Mel频率倒谱系数为特征值的识别系统与HMM/GMM混合模型的RASTA-FF2(RelAtive SpectrAl-Second-order Frequency Filtering)为特征值的识别系统进行比较,并分别在纯语音和带噪语音条件下进行测试,得出:经过二阶频率滤波后的FF2特征值再经过RASTA滤波器滤波,特别是在非稳定噪声环境下,以RASTA-FF2为特征值的识别系统比传统的识别系统的识别率更高.这表明FF2特征值与RASTA滤波器技术相结合,一个作用于频域,一个作用于时间域,可以有效地消除语音信号中的不同噪声成份.

Speech Recognition Based on RASTA-FF2 Filters Denoising Technology

Speech recognition can provide a convenient equipment control means for the people with physical disabilities or the operators who must use two hands simultaneously.In this paper,a robust speech recognition method is introduced in robotic hospital bed control system to enhance robustness in noisy conditions.A combination of the second-order frequency filtering(FF2) with the relative spectral(RASTA) technique for the robust speech recognition system is proposed.The experiments of comparing the traditional HMM/GMM(HMM/Gaussian mixture models) based MFCCs(Mel-frequency cepstral coefficients) recognition system with the HMM/GMM based RASTA-FF2 recognition system were carried out in the conditions of clean and noisy speech respectively.The experimental results show that the new recognition system with RASTA-FF2 features is superior to the traditional one,especially in less stationary noise conditions.This suggests that FF2 combining with Rasta filtering technique may cancel out different noise components in the speech signal by working in the frequency domain and time domain respectively.