Speaker Identification from Mixture of Speech and Non-speech Audio Signal

摘要

Separating speaker from an amalgam of multiple sounds is a challenging area in the domain of speech processing. Henceforth, it has been quickly led to the new area of development in the subfield of speech processing called speaker identification. The proposed work presents a new approach to catch this problem by using acoustic features of the audio signal. The mixture of speech and non-speech audio signal has got separated by using filtering algorithm followed by the recognition of the speech audio by extracting noteworthy acoustic features. A new feature has got implemented as part of contribution to the proposed work named del-MFCC. The computed features have been served for identification of speakers using different popular classifiers. The performance of the presented methodology has been compared with the existing related methods to express the usefulness of the proposed method.