Font Size: a A A

Application Of Mathematical Morphology In Speech Signal Processing

Posted on:2007-09-06Degree:MasterType:Thesis
Country:ChinaCandidate:J L WangFull Text:PDF
GTID:2178360182485311Subject:Physical Electronics
Abstract/Summary:PDF Full Text Request
The speech signal is actually a complicated non-linear non-stationary processes. Recent research effort has started to migrate to analyze speech signal using non-linear theory. These newly developing non-linear theories make up for the imperfections of the traditional linear techniques. The major goal of speech enhancement is extraction as far as possible pure original speech sound from the noisy signal, and speech enhancement is used in speech recognition as a kind of very effective method. Speech recognition is a hotspot in the artificial intelligence. It is popularly used in many fields, which improves human-computer interaction and bring great benefits to us. The performance of most speech recognition systems degrades rapidly in adverse environments that are inevitable in real-word applications.This paper focuses on the research of mathematical morphology in one-dimensional speech signal processing. Based on the traits of digital speech signals, a new morphological filtering algorithm is proposed in speech enhancement. Furthermore, an new algorithm for speech enhancement based on Mathematical Morphology and Wavelets Transform theory. Experiments show the efficiency of removing noisy signal in noisy speech signal by combining mathematical morphology and wavelet transform. In this thesis, we also apply mathematical morphology to sub-band speech signal processing, and present a sub-band speech enhancement method based on mathematical morphology. Morphological filter is used in this method to remove noise in sub-band speech signal, and speech is extracted as pure as possible. A new speech recognition method is designed in the end of this thesis. It this method, noisy speech is pre-processed with mathematical morphology and wavelet transform, and original speech signal is extracted as pure as possible. Then sub-band feature extraction is implemented to pre-processed speech signal, that is to implement speech recognition to pre-processed speech signal by combining sub-band features. The results show that speech recognition rate can be enhanced by implementing pre-processing to noisy speech signal, and the speech recognition effect of sub-band is better than full-band in a way.
Keywords/Search Tags:speech enhancement, speech recognition, sub-band speech, Mathematical Morphology, Wavelet Transform
PDF Full Text Request
Related items