Recognition Influence of Different Acousitc Characters between Male and Female Speakers
Liying Zhao, Qiuyan Wang, Xin Dang · Proceedings of the 2018 2nd International Conference on Computer Science and Artificial Intelligence · 2018
The extraction of speech features is a key technology in the voiceprint recognition system. Correct rate of distinguishing identity of speakers is an important indicator to estimate a feature extraction method. In this paper, we firstly discuss the impact of different phonetic features on speaker recognition between different sexes. Steps and algorithms for obtaining Mel frequency cepstrum coefficient (MFCC), linear predictive cepstral coefficient (LPCC) and two improved MFCC speech features are also introduced in detail. Based on the experimental data presented in this paper, the above mentioned speech features are analyzed and compared theoretically. Experimental results show that recognition ability of male voice is better or worse than that of female voice, according to the feature coefficient is proposed by LPCC algorithm or by MFCC and its improved algorithms, respectively.