Published January 1992
| Published
Journal Article
Open
Speaker-Independent Digit Recognition Using a Neural Network with Time-Delayed Connections
Abstract
The capability of a small neural network to perform speaker-independent recognition of spoken digits in connected speech has been investigated. The network uses time delays to organize rapidly changing outputs of symbol detectors over the time scale of a word. The network is data driven and unclocked. To achieve useful accuracy in a speaker-independent setting, many new ideas and procedures were developed. These include improving the feature detectors, self-recognition of word ends, reduction in network size, and dividing speakers into natural classes. Quantitative experiments based on Texas Instruments (TI) digit databases are described.
Additional Information
© 1992 Massachusetts Institute of Technology. Received 11 February 1991; accepted 15 July 1991. The TI connected digit data base was provided by the National Bureau of Standards. We wish to thank David Talkin for providing us the WAVES program and the Speech Research Department at Bell Labs for computer support. The work of J.J.H. at Caltech was supported in part by Office of Naval Research (Contract No. N00014-87-K-0377).Attached Files
Published - UNNnc92.pdf
Files
UNNnc92.pdf
Files
(717.4 kB)
Name | Size | Download all |
---|---|---|
md5:e060321777093cee589acbca4c2f3a46
|
717.4 kB | Preview Download |
Additional details
- Eprint ID
- 12332
- Resolver ID
- CaltechAUTHORS:UNNnc92
- Office of Naval Research
- N00014-87-K-0377
- Created
-
2008-11-12Created from EPrint's datestamp field
- Updated
-
2021-11-08Created from EPrint's last_modified field