APPARATUS FOR PREPROCESSING OF SPEECH SIGNAL AND METHOD FOR EXTRACTING END-POINT OF SPEECH SIGNAL THEREOF

摘要

An apparatus and a method for preprocessing a voice signal are provided to extract a characteristic of a noise frame different from a voice and improve the processing performance of the voice signal. A noise and voice judging unit(150) discriminates a noise and a voice on the basis of at least one of an initial noise frame and an end noise frame. The noise and voice judging unit includes an initial or end noise frame calculator(151), an SNR(Signal to Noise Ratio) calculator(153), a noise information update device(155), and a noise judging device(157). The initial or end noise frame calculator calculates noise information by using energy of the initial or end noise frame. The SNR calculator calculates an energy ratio of a voice signal and a noise. The noise information update device updates and stores noise information of the initial or end noise frame and noise information of a frame judged as a noise frame in the noise judging device. The noise judging device compares an SNR of a current frame calculated in the SNR calculator with the noise information stored in the noise information update device.