Audio acoustic features-based instrument recognition using classification algorithms
by Anuja Arora; Raghav Pangasa; Somya Goel; Tribhuwan Kumar Tewari
International Journal of Advanced Intelligence Paradigms (IJAIP), Vol. 20, No. 1/2, 2021

Abstract: Musicians have strong knowledge to identify instruments from audio and they can easily categorise music samples by instruments. While on the contrary, instrument recognition is a nascent problem of machine perception area. In this research work, a comparative study of various classification models in order to recognise instrument is presented. Instruments are recognised and classified in the audio on the basis of audio acoustic features. Four machine learning classification algorithms - support vector machine, decision tree, random forest, and ensemble models are applied to classify and tag instruments in audio files based on audio acoustic features. Two well-known datasets, IRMAS and NSynth, are used to apply various classification models and to validate the role of audio acoustic features in instrument recognition. Instrument recognition result shows that the IRMAS dataset achieves maximum accuracy of 74.60% using ensemble model whereas the NSynth dataset gains 96.89% maximum accuracy using the random forest model.

Online publication date: Thu, 16-Sep-2021

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Advanced Intelligence Paradigms (IJAIP):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com