Time–Frequency Feature Fusion for Noise Robust Audio Event Classification
- Submitting institution
-
The University of Kent
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 16752
- Type
- D - Journal article
- DOI
-
10.1007/s00034-019-01203-0
- Title of journal
- Circuits Systems and Signal Processing
- Article number
- -
- First page
- 1672
- Volume
- 39
- Issue
- 3
- ISSN
- 0278-081X
- Open access status
- Compliant
- Month of publication
- July
- Year of publication
- 2019
- URL
-
https://kar.kent.ac.uk/75276/
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
4
- Research group(s)
-
-
- Citation count
- 1
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- This is the first paper to merge a biologically-inspired hearing model (cochleogram) with spectrogram features for sound event detection (SED), and also the first to evaluate a constant-Q transform for the task. This work is significant because the method improved state-of-the-art SED (for 0 dB SNR) from 85% to over 97% and was additionally shown suitable for deployment in noisy real-world settings.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
