Abstract
The idea of extracting typical Mel-frequency cepstral coefficients (MFCC) conducts efficient improvement in speech signal processing, though these coefficients only apply the information of magnitude without phase. This article’s purpose is to form a sort of formant instantaneous characteristics (FIC) using phase information and to see these phase parameters’ function for speaker identification (SI). The procedure requests applications of Hilbert Transform (HT), band-pass filters, and mel-frequency perceptual warping. FIC together with MFCC were tested in SI experiments based on a Gaussian mixture model (GMM). And results show that FIC play a compensating role to MFCC in SI, with one of improved relative rate up to 10.13%. Experimental utterances are Chinese mandarin under clean recording circumstances.