@article { author = {Morovati, Mehdi}, title = {Increase the accuracy of speech signal categories in high noise environments}, journal = {Journal of Bioengineering Research}, volume = {2}, number = {3}, pages = {1-13}, year = {2020}, publisher = {Tissues and Biomaterial Research Group-(TBRG)}, issn = {2645-5633}, eissn = {}, doi = {10.22034/jbr.2020.232801.1023}, abstract = {In this paper, in order to improve the quality and reconstruction of the noisy speech signal, an algorithm based on clustering in the spectrographic image texture is proposed and evaluated. In this algorithm, we first determine the cluster centers by the method of Glass Compressed Transformation (GCT), based on image tissue extraction and Weighted K-mean algorithm in the field of GCT. After determining the centers of the clusters and their variance, by using the Gaussian Distribution Function (GDF), we determine the mask for each cluster in the GCT space and apply a signal to the GCT domain. In the continuation of this research, this method can be used to separate the two speakers that have been combined and it also can be used in separating speakers in noisy situations. The results of the method evaluation in the presence of a variety of conventional acoustic noises show the advantages of this method in improving the quality of speech.}, keywords = {Weighted K-mean,Gaussian Distribution Function,Clustering,Mask}, url = {https://www.journalbe.com/article_109406.html}, eprint = {https://www.journalbe.com/article_109406_e567d23b0039daaec619d6be4245975a.pdf} }