The perceptual wavelet feature for noise robust Vietnamese speech recognition

This study proposes a novel noise robust speech feature for Vietnamese speech recognition based on the Bark scale, the Perceptual Wavelet Packet Transform (PWPT), the Wavelet sub-band Energy Parameters (WEP). This method is used in compare to the widely used MFCC feature recognizer. The experimental results show that our proposal is superior to the widely used MFCC feature for Vietnamese speech corpus with both clean and white noisy speech.