Abstract
convolutional neural network (CNN) which is a class of deep, feed-forward artificial neural network. We decided to analyze audio signal feature maps, namely spectrograms, linear and Mel-scale cepstrograms, and chromagrams. The choice was made upon the fact that CNN performs well in 2D data-oriented processing contexts. Feature maps were employed in the Lithuanian word recognition task. The spectral analysis led to the highest word recognition rate. Spectral and mel-scale cepstral feature spaces outperform linear cepstra and chroma. The 111-word classification experiment depicts f1 score of 0.99 for spectrum, 0.91 for mel-scale cepstrum , 0.76 for chromagram, and 0.64 for cepstrum feature space on test data set.
Citations
-
2 7
CrossRef
-
0
Web of Science
-
3 9
Scopus
Authors (5)
Cite as
Full text
full text is not available in portal
Keywords
Details
- Category:
- Articles
- Type:
- artykuł w czasopiśmie wyróżnionym w JCR
- Published in:
-
JOURNAL OF THE AUDIO ENGINEERING SOCIETY
no. 66,
pages 1072 - 1081,
ISSN: 1549-4950 - Language:
- English
- Publication year:
- 2018
- Bibliographic description:
- Korvel G., Treigys P., Tamulevicus G., Bernataviciene J., Kostek B.: Analysis of 2D Feature Spaces for Deep Learning-based Speech Recognition// JOURNAL OF THE AUDIO ENGINEERING SOCIETY. -Vol. 66, nr. 12 (2018), s.1072-1081
- DOI:
- Digital Object Identifier (open in new tab) 10.17743/jaes.2018.0066
- Sources of funding:
-
- Statutory activity/subsidy
- Project Methodology and technology for the polymodal allophonic speech transcription
- Verified by:
- Gdańsk University of Technology
seen 223 times
Recommended for you
Investigating Feature Spaces for Isolated Word Recognition
- G. Korvel,
- G. Tamulevicus,
- P. Treigys
- + 2 authors
Investigating Feature Spaces for Isolated Word Recognition
- P. Treigys,
- G. Korvel,
- G. Tamulevicius
- + 2 authors