The 5-Second Challenge: Detection of Derivative Works from Original Music using Acoustic Cues

View at publisher Summarise with VIRAT
Authors: Sanjeev Kumar Mittal, Rama Krishna Challa, Mukesh Kumar Rohil
Cited by: 0
References: 31

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

With the advent of Artificial Intelligence, online mimicry and fake narratives are on the rise. Various past studies have tried to detect Derivative Works using pattern analysis and Natural Language Processing based on the features extracted using the Speech Signal Processing. This study presents a light weight and an efficient mechanism that is based on only the Acoustic Features to detect the Parody from the Original Song that are clips of under 5 seconds of duration. The dataset is collected as short videos that are parodies of the popular Hindi Songs and are downloaded from YouTube Channels. Results with 89.29% accuracy and F1 score of 0.9032 were achieved by using all the Mel Frequency Cepstral Coefficients (MFCC) and Chroma features with the K-Nearest Neighbors (KNN) classifier. This study is useful in advancing the field of deception and deceit detection using the latest Machine Learning and Signal Processing techniques.

Fields of Study & Subjects

Music and Audio Processing Speech Recognition and Synthesis Digital Media Forensic Detection

Publication details

Journal / source
International Conference on Signal Processing and Electronics Design (ICSPED)
Pages
37–42
Publication date
23 January 2026
Language
EN