Font Size: a A A

Implemetation Of Sound Source Localization Method In Low Snr And Reverberant Environment

Posted on:2015-10-17Degree:MasterType:Thesis
Country:ChinaCandidate:Y Z WeiFull Text:PDF
GTID:2308330473953958Subject:Electronic and communication engineering
Abstract/Summary:PDF Full Text Request
The accuracy of sound source localization system would be affected by reverberation and low SNR in tight spaces environment, which might even cause the system cannot work properly. Reverberation cannot be filtered out by routine filters.Thus, it was necessary to propose a proved sound source localization method to filter out noise and reverberation.Against the above problems, aiming at the environment of low SNR noise and reverberation, this thesis took T-style microphone array for example, proposed a method based on mathematical morphology. The main contents were as follows:Firstly, to order to solve the problem of the low frequency resolution using wavelet filter alone and poor effect when handling the Hopping signal, this thesis innovatively put forward a type of improved filter combining morphological filter and wavelet filter.Take white noise, destroyer-engine noise and f16 noise for instance, morphological filter had been used to higher the low SNR signal firstly, then wavelet filter had been used to deal with the noise to achieve the goal of noise filtering.Secondly, based on process above, this thesis combined the wavelet packet preprocessing methods and empirical mode decomposition method to separate reverberation signal, and then reconstructed the original signal. Take the speech adding reverberation signal for example, the signal can be divided into a series of narrow-band signal by using wavelet packet preprocessing, the empirical mode decomposition had been used to obtain a series of signals characterizing the original signal, then via filtering and reconstructing the signal, eventually pure signal could be got.Finally, basing the sound source localization algorithm based on time delay estimation, this thesis designed and implemented sound source localization systems.According to the experimental verification, the proposed method based on mathematical morphology could filter out the noise and reverberation effectively, laid the foundation for further improving accuracy, enhanced the immunity of external environmental factors.
Keywords/Search Tags:Sound Source Localization, Mathematical Morphology, EMD, IMF, TDE
PDF Full Text Request
Related items