• AMRUTHA K V
• HEMA KRISHNAN

Audio sentiment analysis utilizing automatic speech recognition is a rising research area where opinion or sentiment shown by a speaker is distinguished from normal sound. It is generally underexplored when contrasted with text based sentiment detection. Extracting speaker sentiment from nat- ural audio sources is a challenging problem. Conventional techniques for sentiment extraction generally use transcripts from a speech recognition system, and process the transcript utilizing text based sentiment classifiers. This paper proposes a text based sentiment classifier which decides the most valuable and discriminative sentiment bearing keyword terms, utilized as a term list for KWS. To get a conservative yet discriminative sentiment term list, iterative feature optimization for maxi- mum entropy sentiment model is proposed to reduce model complexity while keeping up effective classification accuracy. A new hybrid ME-KWS joint scoring methodology is created to display both text and audio based parameters in a single integrated formulation. After this, a comparison study was conducted with SVM classifier which indicates moderately poor accuracy. Test results demonstrate that the proposed KWS based framework fundamentally beats the customary ASR architecture in detecting sentiment for challenging practical undertakings.

"AUTOMATIC SENTIMENT DETECTION IN TEXT AND NATURALISTIC AUDIO", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 6, page no.581-586, June-2019, Available :http://www.jetir.org/papers/JETIR1906229.pdf

"AUTOMATIC SENTIMENT DETECTION IN TEXT AND NATURALISTIC AUDIO", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 6, page no. pp581-586, June-2019, Available at : http://www.jetir.org/papers/JETIR1906229.pdf