Speaker Change Detection for News Audio: A Practical Approach

  MsUzma Muzzaffar,  Dimple Goyal,  Muheet Ahmed Butt

Speaker change detection is imperative approach for automatic division process when various speakers communicate at a same case of time. The speech information is decayed into different homogeneous sections with each segment contains the audio information of a single speaker. Existing methodologies for speaker change detection complete disparity of the dispersion of the audio information in a computerized frame prior and then afterward a speaker change point. In this paper, we propose speaker segmentation and clustering based strategy for speaker change identification. Speaker diarisation (or segmentation) is the way toward dividing an information sound stream into homogeneous sections as per the speaker character. Speaker diarisation is a mix of speaker division and speaker grouping. Highlights removed from the sound information around the speaker change focuses are utilized as positive cases. Highlights separated from the information between the speaker changes focuses are utilized as negative illustrations. The positive and negative illustrations are utilized as a part of preparing a help vector machine for speaker change detection.

Keywords- Speaker Segmentation and Change Detection for News Audio: A Practical Implementation

Unique Identification Number - IJEDR1704071
Page Number(s) - 457-462
Pubished in - Volume 5 | Issue 4 | November 2017
DOI (Digital Object Identifier) -   
Publisher - IJEDR (ISSN - 2321-9939)

  MsUzma Muzzaffar,  Dimple Goyal,  Muheet Ahmed Butt,   "Speaker Change Detection for News Audio: A Practical Approach", International Journal of Engineering Development and Research (IJEDR), ISSN:2321-9939, Volume.5, Issue 4, pp.457-462, November 2017, Available at :http://www.ijedr.org/papers/IJEDR1704071.pdf