Speaker Identification for Swiss German with Spectral and Rhythm Features

We present results of speech rhythm analysis for automatic speaker identification. We expand previous experiments using similar methods for language identification. Features describing the rhythmic properties of salient changes in signal components are extracted and used in an speaker identification task to determine to which extent they are descriptive of speaker variability. We also test the performance of state-of-the-art but simple-to-extract frame-based features. The paper focus is the evaluation on one corpus (swiss german, TEVOID) using support vector machines. Results suggest that the general spectral features can provide very good performance on this dataset, whereas the rhythm features are not as successful in the task, indicating either the lack of suitability for this task or the dataset specificity. DOI: https://doi.org/10.17743/aesconf.2017.978-1-942220-15-2 Posted at the Zurich Open Repository and Archive, University of Zurich ZORA URL: https://doi.org/10.5167/uzh-138208 Published Version Originally published at: Lykartsis, Athanasios; Weinzierl, Stefan; Dellwo, Volker (2017). Speaker Identification for Swiss German with Spectral and Rhythm Features. In: 2017 AES International Conference on Semantic Audio (June 2017), Erlangen, June 2017 June 2017. DOI: https://doi.org/10.17743/aesconf.2017.978-1-942220-15-2 Audio Engineering Society Conference Paper Presented at the Conference on Semantic Audio 2017 June 22 – 24, Erlangen, Germany This paper was peer-reviewed as a complete manuscript for presentation at this conference. This paper is available in the AES E-Library (http://www.aes.org/e-lib) all rights reserved. Reproduction of this paper, or any portion thereof, is not permitted without direct permission from the Journal of the Audio Engineering Society. Speaker Identification for Swiss German with Spectral and Rhythm Features Athanasios Lykartsis1, Stefan Weinzierl1, and Volker Dellwo2 1Audio Communication Group, Technische Universität Berlin, Germany 2Phonetics Laboratory, Universität Zürich, Switzerland Correspondence should be addressed to Athanasios Lykartsis ([email protected]) ABSTRACT We present results of speech rhythm analysis for automatic speaker identification. We expand previous experiments using similar methods for language identification. Features describing the rhythmic properties of salient changes in signal components are extracted and used in an speaker identification task to determine to which extent they are descriptive of speaker variability. We also test the performance of state-of-the-art but simple-to-extract frame-based features. The paper focus is the evaluation on one corpus (swiss german, TEVOID) using support vector machines. Results suggest that the general spectral features can provide very good performance on this dataset, whereas the rhythm features are not as successful in the task, indicating either the lack of suitability for this task or the dataset specificity.We present results of speech rhythm analysis for automatic speaker identification. We expand previous experiments using similar methods for language identification. Features describing the rhythmic properties of salient changes in signal components are extracted and used in an speaker identification task to determine to which extent they are descriptive of speaker variability. We also test the performance of state-of-the-art but simple-to-extract frame-based features. The paper focus is the evaluation on one corpus (swiss german, TEVOID) using support vector machines. Results suggest that the general spectral features can provide very good performance on this dataset, whereas the rhythm features are not as successful in the task, indicating either the lack of suitability for this task or the dataset specificity.