Volume 45, Issue 2, April 2025, Pages 439–447
N’Tcho Assoukpou Jean GNAMELE1, Bi Tra Jean Claude YOUAN2, and Konan Fernand GBAMELE3
1 Department of Mathematics-Physics-Chemistry at University Peleforo Gon Coulibaly (UPGC), BP 1328 Korhogo, Côte d’Ivoire
2 Technology Laboratory, Félix Houphouët-Boigny University (UFHB), 01 BP V34 Abidjan 01 Cocody, Côte d’Ivoire
3 Optical Networks and Telecommunications Laboratory of the Joint Research and Innovation Unit in Engineering Science and Technology (UMRI-STI), National Polytechnic Institute Félix Houphouët-Boigny (INPHB), Bp 1093 Yamoussoukro, Côte d’Ivoire
Original language: English
Copyright © 2025 ISSR Journals. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This paper investigates the relevance of using a large number of Mel Frequency Cepstral Coefficients (MFCC) as descriptors of acoustic signals, and the interaction between these and the nature of the frequency band in which the Mel filters are arranged. This study forms part of the wider field of automatic recognition of acoustic signals, with a particular focus on those that are not speech-related. We evaluated a series of MFCCs, spanning a range from 1 to 50, utilising the central octave band frequencies (31.5 Hz-16000 Hz) as the MFCC calculation frequencies. An application was made to the identification of chainsaw sounds among a plurality of signals from the forest environment. The results revealed a threshold value for the number of MFCCs (LVMFCC) above which classification rates remain constant. The LVMFCC=39 was common to all frequencies, although specifically the LVMFCC for each centre frequency was between 5 and 39 MFCCs. We observed that the notion of an optimal value for the number of MFCCs could appear subjective. The best classification rate of 98.41% obtained with the 16000 Hz centre frequency corresponds to a number of MFCCs between 5 and 50. These results also reveal the need to restructure the.
Author Keywords: acoustic, automatic recognition, KNN, MFCC, octave band.
N’Tcho Assoukpou Jean GNAMELE1, Bi Tra Jean Claude YOUAN2, and Konan Fernand GBAMELE3
1 Department of Mathematics-Physics-Chemistry at University Peleforo Gon Coulibaly (UPGC), BP 1328 Korhogo, Côte d’Ivoire
2 Technology Laboratory, Félix Houphouët-Boigny University (UFHB), 01 BP V34 Abidjan 01 Cocody, Côte d’Ivoire
3 Optical Networks and Telecommunications Laboratory of the Joint Research and Innovation Unit in Engineering Science and Technology (UMRI-STI), National Polytechnic Institute Félix Houphouët-Boigny (INPHB), Bp 1093 Yamoussoukro, Côte d’Ivoire
Original language: English
Copyright © 2025 ISSR Journals. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract
This paper investigates the relevance of using a large number of Mel Frequency Cepstral Coefficients (MFCC) as descriptors of acoustic signals, and the interaction between these and the nature of the frequency band in which the Mel filters are arranged. This study forms part of the wider field of automatic recognition of acoustic signals, with a particular focus on those that are not speech-related. We evaluated a series of MFCCs, spanning a range from 1 to 50, utilising the central octave band frequencies (31.5 Hz-16000 Hz) as the MFCC calculation frequencies. An application was made to the identification of chainsaw sounds among a plurality of signals from the forest environment. The results revealed a threshold value for the number of MFCCs (LVMFCC) above which classification rates remain constant. The LVMFCC=39 was common to all frequencies, although specifically the LVMFCC for each centre frequency was between 5 and 39 MFCCs. We observed that the notion of an optimal value for the number of MFCCs could appear subjective. The best classification rate of 98.41% obtained with the 16000 Hz centre frequency corresponds to a number of MFCCs between 5 and 50. These results also reveal the need to restructure the.
Author Keywords: acoustic, automatic recognition, KNN, MFCC, octave band.
How to Cite this Article
N’Tcho Assoukpou Jean GNAMELE, Bi Tra Jean Claude YOUAN, and Konan Fernand GBAMELE, “MFCC number limit for automatic sound recognition: Application to the chainsaw sound identification,” International Journal of Innovation and Applied Studies, vol. 45, no. 2, pp. 439–447, April 2025.
