A Frequency-Weighted Itakura Spectral Distortion Measure and Its Application to Speech Recognition in Noise

The Itakura LPC spectral distortion measure has been applied successfully to the recognition of noise-free speech signals. However, the performance of a recognizer based on this distortion measure deteriorates when the speech signal is corrupted by noise. In certain applications, where training and testing can be done under the same noise conditions, the degradation in performance is often small. For other applications, where it is not feasible to train and test the recognizer in similar noise environments, the performance is often unacceptable. In such situations a more noise-tolerant distortion measure is required. Since, in general, the speech power is nonuniformly distributed along the frequency axis, we consider a weighted spectral distortion measure which weights the high SNR regions more than the low SNR regions in frequency.