Speech under Stress: Analysis and Recognition using KNN

Speech under Stress: Analysis and Recognition using KNN

Bhawna Sharma1

bhawna190sharma@gmail.com

1Assistant Professor, Guru Nanak Khalsa College, Yamunanagar

ABSTRACT

This work examines how speech characteristics might be analyzed to tailor automated conversation recognition to illness diagnosis. Voice analysis, which is now performed by a skilled physician using techniques based on auditory analysis, enables the diagnosis of illnesses that impact the vocal apparatus. The suggested work offers a cutting-edge way to monitor a patient's pathology. It is stress-free to use, quick, and affordable for the physician, and safe for the patient. This technique practices a parametric methodology (jitter, shimmer, harmonic to sound, etc.) to assess the sick vocal sound. Additionally, the technique for this work rest on on the feature extraction of Mel Frequency Cepstral Coefficient (MFCC) and the feature matching of Dynamic Time Warping (DTW). This article provides a step-by-step breakdown of the speech analysis procedure used on anxious individuals. The degree of classification accuracy attained for each of the retrieved characteristics confirms that our method is effective in differentiating between stressed individuals. Thus, the speech analysis feature on the PRAAT platform has produced positive results and shown to be an effective tool for differentiating between those who are under stress and those who are not. Using MATLAB 2018b, the work displays several 2D and 3D displays of teaching and anticipated matters.

Keywords: Dynamic Time Warping, Mel Frequency Cepstral Coefficient, K nearest neighbor, Virtual Reality