Abstract | ||
---|---|---|
A novel type of feature extraction for automatic speech recog- nition is investigated. Two-dimensional Gabor functions, with varying extents and tuned to different rates and di- rections of spectro-temporal modulation, are applied as fil- ters to a spectro-temporal representation provided by mel spectra. The use of these functions is motivated by find- ings in neurophysiology and psychoacoustics. Data-driven parameter selection was used to obtain Gabor feature sets, the performance of which is evaluated on the Aurora 2 and 3 datasets both on their own and in combination with the Qualcomm-OGI-ICSI Aurora proposal. The Gabor features consistently provide performance improvements. |
Year | Venue | Keywords |
---|---|---|
2002 | INTERSPEECH | feature extraction |
Field | DocType | Citations |
Psychoacoustics,Pattern recognition,Neurophysiology,Computer science,Feature extraction,Speech recognition,Artificial intelligence,Kanade–Lucas–Tomasi feature tracker,Word accuracy | Conference | 37 |
PageRank | References | Authors |
2.50 | 4 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Michael Kleinschmidt | 1 | 95 | 6.48 |
michael kleinschmidt a b | 2 | 37 | 2.50 |
David Gelbart | 3 | 134 | 17.54 |