We describe in this paper a speaker independent, global word recognition task using time delay networks. We first describe these networks as a way for learning feature extractors by constrained back-propagation. Such a time-delay network is shown to be capable of dealing with a test task: French digit recognition. The results are discussed and compared, on the same data sets, with those obtained with a classical time warping system. Both connectionist and classical systems achieved no more than 1% errors on the test set.
@inproceedings{bottou89_eurospeech, title = {Experiments with time delay networks and dynamic time warping for speaker independent isolated digits recognition}, author = {L. Bottou and F. {Fogelman Soulie} and Pascal Blanchet and Jean-Sylvain Lienard}, year = {1989}, booktitle = {First European Conference on Speech Communication and Technology (Eurospeech 1989)}, pages = {2537--2540}, doi = {10.21437/Eurospeech.1989-91}, issn = {1018-4074},}Cite as:Bottou, L., Fogelman Soulie, F., Blanchet, P., Lienard, J.-S. (1989) Experiments with time delay networks and dynamic time warping for speaker independent isolated digits recognition. Proc. First European Conference on Speech Communication and Technology (Eurospeech 1989), 2537-2540, doi: 10.21437/Eurospeech.1989-91