A Research On Text-independent Speaker Recognition

Speaker recognition researches commenced at about 1930. This field has become more and more important since 1970' s. The technique of speaker recognition can be applied to many applications, such as speaker checking, judicial evidence validation, medical applications, voice-controlled entrance permission, voice-controlled account access etc. Due to its huge potential market, many institutes and researchers have been involved in the research many years and acquired great achievements. However it is still far to a complete success.This paper is mainly focused on the research of the text-independent speaker recognition technique. Firstly, the fundamentals of the speaker recognition are discussed in detail. Then the followings are preprocessing and feature extraction procedures. After that are the windowing, noise filtering, end-points detecting of speech signals. The double threshold end-points detecting is discussed in detail ,following a improved end-points detecting : based on ICA enhancement and spectral entropy. Still then, MFCC, the most important speech feature extracting methods for speaker recognition, is intensively investigated and is chosen as feature extractor for speech signals.In addition, several speaker recognition algorithms, including the VQ and GMM, are thoroughly investigated. For VQ algorithm, the definition of VQ concept, distortion measure, and the method for fittest codebook design are firstly fully unwrapped, then delved into fuzzy VQ(FVQ). Experiments showed that FVQ improved correct recognition rate about 10% than VQ. For GMM algorithm, GMM concept, principles of modeling and estimation theory for model parameters are completely uncovered, several experiments are also conducted. All these experiments have shown, under the same experimental conditions, GMM algorithm secured the best recognition result.In the final, conclusion has been derived about the research, and some advice for further researches has also been provided.