The classification of emotion based on human voice by using Mel Frequency Cepstrum Coefficient (MFCC) and Naive Bayes method

Emotion is a manifestation of the emotions of an individual resulting from such occurrences. Emotions will, in general, be classified as anger, fear, sadness and pleasure. A structured and logical mechanism is needed for the computer to carry out the classification process. The Mel Frequency Coeffic...

Full description

Saved in:
Bibliographic Details
Main Authors: Safriadi, Mahlil, Hidayat, Hari Toha, Nasir, Muhammad, Anwar
Format: Conference Proceeding
Language:English
Subjects:
Classification
Coefficients
Emotions
Training
Voice recorders
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Emotion is a manifestation of the emotions of an individual resulting from such occurrences. Emotions will, in general, be classified as anger, fear, sadness and pleasure. A structured and logical mechanism is needed for the computer to carry out the classification process. The Mel Frequency Coefficient (MFCC) approach plays a crucial role in extracting and representing features from sounds in numerical data. The Naïve Bayes method is a classification method that expects the Bayes theory of classification. In the research, the training voice data used were 100 voice data and the testing data used 100 voice data. The voice data used in the training process and the testing process for emotional classification are voice data recorded with a voice recorder * wav format. The recording duration is 15 seconds. The classification of emotions based on sound using the Mel Frequency Coefficient (MFCC) and Naive Bayes method were 72% were successfully classified correctly, and 28% were not classified correctly. The emotional classification performance rate is highly impacted by the amount of training data used and the noise from the training data and data testing.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0117958