Algorithm Selection for Automated Audio Classification based on Content

Draganov, Ivo and Minchev, Krasimir (2019) Algorithm Selection for Automated Audio Classification based on Content. In: 54th International Scientific Conference on Information, Communication and Energy Systems and Technologies (ICEST 2019), June 27-29, 2019, Ohrid, North Macedonia.

[thumbnail of ICEST2019_25.pdf]

Download (995kB) | Preview


In this paper an approach for algorithm selection for automated audio classification is proposed based on content. Three popular algorithms for speech vs. music discrimination are incorporated, namely the zero-crossing rate, average frame power and 4-Hz energy modulation which prove effective at different level depending on the content of the audio being processed. Extensive testing is done over various musical pieces from different countries and in different styles to evaluate the accuracy and time consumption for each one. Then based on
registered levels and with statistics from the initial record for particular audio recording the most proper could be switched on for processing in a complete audio classification system. Obtained results re considered promising for future use on a wider scale.

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: Audio Classification, Music Speech Discrimination, 4 Hz Energy Modulation, Zero-Crossing Rate, Average Frame Power
Subjects: Scientific Fields (Frascati) > Engineering and Technology > Electrical engineering, electronic engineering,information engineering
Divisions: Faculty of Technical Sciences
Depositing User: ICEST
Date Deposited: 18 Mar 2020 17:07
Last Modified: 18 Mar 2020 17:07

Actions (login required)

View Item View Item