Tomaszewska, Julia Z., Chousidis, C. and Georgakis, Apostolos (2024) COMPARATIVE ANALYSIS OF MFCC AND GTCC PERFORMANCE IN LARYNGEAL PATHOLOGY DETECTION BASED ON ELECTROGLOTTOGRAPHIC SIGNALS. In: Acoustics 2024, 12-13 Sep 2024, Manchester Metropolitan University, Manchester, UK.

Preview

PDF JZ TOMASZEWSKA C CHOUSIDIS A GEORGAKIS COMPARATIVE ANALYSIS OF MFCC AND GTCC PERFORMANCE IN LARYNGEAL PATHOLOGY DETECTION BASED ON ELECTROGLOTTOGRAPHIC SIGNALS.pdf - Published Version Download (227kB) | Preview

Abstract

In the following paper, we analyse and compare the performance of Mel-Frequency Cepstral Coefficients (MFCC) and Gammatone Cepstral Coefficients (GTCC) in recognising the pathological patterns related to laryngeal disorders in electroglottographic signals. Furthermore, we investigate and compare the performance of two data types in laryngeal pathology detection; the bio-impedance measurements of sustained phonation and bio-impedance signals obtained during continuous speech. The ability of GTCC and MFCC to recognise pathological patterns in both types of bioimpedance signals is assessed using the designed CNN classifier. For both data types, the obtained results demonstrated that the GTCCs are superior in recognising pathological patterns in bioimpedance signals than MFCCs. Moreover, the speech data outperforms sustained phonation in detection of laryngeal pathologies. The achieved accuracy of the proposed CNN system with the application of the MFCCs derived from sustained phonation delivered 88.69% ±3.14 accuracy. In contrast, the proposed system fed GTCCs derived from speech delivered 95.95% ±1.25 accuracy.

Item Type:	Conference or Workshop Item (Paper)
ISBN:	9781906913496
Identifier:	10.25144/23671
Identifier:	10.25144/23671
Subjects:	Construction and engineering > Biomedical engineering
Related URLs:	Publisher
Depositing User:	Marc Forster
Date Deposited:	20 Jan 2025 09:20
Last Modified:	20 Jan 2025 09:30
URI:	https://repository.uwl.ac.uk/id/eprint/13105

Downloads

Actions (login required)

View Item