Font Size: a A A

The Realization Of Audio Discrimination System Based On VQ

Posted on:2011-05-30Degree:MasterType:Thesis
Country:ChinaCandidate:H P JiangFull Text:PDF
GTID:2178330332471450Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
With the development of Internet technology, Because of the vast data increments on various types of date, how to find out the audio types we need from the vast date is concerned more and more, as a result audio data analysis technology comes up, The audio data analysis technology has great plays an important role about the machine sense the surroundings automatically. Because over 80% information received from visual and auditory, as the audio information has the feature of intuitive, to solve this problem from the audio point of view become the ideal means ,through which the problem about a voice and non-speech classification issues belongs the audio scene classification areas.In the process of control voice, there is a problem of that voice signal and non-voice signal parts of noise may be mistaken to voice signal, it cause the misoperation of system. To solve this problem, this article based on the theory of speaker recognition, a method which use vector quantization is presented to confirm the type of audio signal, so that to distinguish the non-speech signal and eliminate it later, advance the system's work efficiency. First, system extracts the mel frequency cepstrum coefficient from the audio waveform sequences that have been gathered, and then obtains feature vector sequence of audio. Next use the algorithm of vector quantization to obtain the template of voice signal and non-voice signal respectively. Last, carry out the distorting measure between the feature vector sequence of audio awaiting recognition and each template, then select the one with smallest quantization error to be the type which is awaiting recognition. Thus, that makes the discrimination and identification of audio signal realized.In this paper we set up an audio discrimination system used to achieve the confirmation of audio type based on matlab the test result indicates that this system was effective for the discrimination of voice signal and non-voice signal, and put forward a plan for DSP, to make this system can be used in practical hardware system.
Keywords/Search Tags:vector quantization, Mel frequency cepstrum coefficient, lbg algonithm, end point detector
PDF Full Text Request
Related items