Music genre predictor based classification of audio files with low level feature of frequency and time domain using support vector machine over multiclass support vector machine.

Using Novel Support Vector Machine (SVM) and Multiclass Support Vector Machine (SVM-MC) for genre-based classification of audio recordings by using low-level information in the frequency and temporal domain is the major purpose of this work (MSVM). In the current investigation, we make use of two case studies: one for the SVM technique, and the other for the MSVM methodology. To obtain the desired values of 80 percent G power, 0.05 percent threshold, and 95 percent CI, the device was sampled ten times, and with each iteration, the accuracy was improved by 5 percent. When it comes to predicting the classification of Audio files using the low-level characteristic of frequency, the SVM approach achieves a greater success rate of 91.25 percent as opposed to the MSVM method, which achieves a success rate of 84.78 percent. When comparing the two groups, a significance level of 0.008 (p0.05) implies that there is a statistically significant gap between them. When it comes to forecasting low-frequency audio data, the SVM approach performs much better than the CNN algorithm. [ABSTRACT FROM AUTHOR]