Melody-Based Hindi Song Retrieval Using SVM

Molakathaala, Naagamani and Gautam, Veerendra Kumar and Vasu, Violin and et al, . (2022) Melody-Based Hindi Song Retrieval Using SVM. In: 5th International Conference on Smart Computing and Informatics, SCI 2021, 17 September 2021through 18 September 2021, Hyderabad.

Full text not available from this repository. (Request a copy)

Abstract

The main purpose of this paper is to write the function for audio information extractions and designing a model for classifying the Hindi songs based on the melody as the pattern. Using the R-programming, we can call the audio analysis tool to extract the basic parameter that is used to represent the melody. By using R-programming, the Praat tool and Praat script define the model. This model will extract features of the songs that represent melody; they are pitch, intensity, formant, mel-frequency cepstrum coefficient (MFCC). These parameters are extracted for each frame which is a portion of the audio song file. In this work, 30 s duration frame was used. Then calculate the number of frames, total duration, mean, standard deviation, minimum (pitch, intensity, etc.), and maximum (pitch, intensity, etc.) with Praat script. The defined script is a function using the R-program for the defined model. In present work nearly 120 Hindi songs are used. All these songs were converted into a frame size of 30 s duration. In this way, dataset for training and test modules is prepared. This model uses the SVM technique and classifies the Hindi songs into different categories like sad, happy, romantic. By preparing this information in a particular order, this model classifies music by analyzing the information of the song. Because nowadays numerous songs are been released, the genre of the song clustering is reasonably more important in terms of the audience’s choice. Also, arguments for plagiarism are continuously being raised. This type of model will help in selection of good music and also for plagiarism. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

[error in script]

IITH Creators: