The Study of The Speaker Testing System Based on The Object-oriented Method

摘要

Speaker Recognition is an important branch of speech signal .In the paper, aiming at the limitations of the research of the current speaker recognition, a speaker recognition software-development platform was constructed by means of the object-oriented method. On the base, the Speaker-recognition experiment was carried out using the testing system, at the same time, the differences between linear predictive cepstral parameters and beauty Seoul cepstral parameters, the HMM model and the GMM model were analyzed and compared based on the experimental results, and a certain reference and reference value will be provided when speaker recognition become practical.