Using sparse representations for exemplar based continuous digit recognition

Abstract

This paper introduces a novel approach to exemplar-based con-nected digit recognition. The approach is tested for different sizes of the exemplar collection (from 250 to 16,000), different length of the exemplars (from 1 to 50 time frames) and state-labeled versus word-labeled decoding. In addition, we compare the novel method for selecting exemplars, based on Sparse Classification, with a con-ventional K-Nearest-Neighbor approach. For word-labeled decod-ing we developed a Viterbi search that applies minimum and maxi-mum duration constraints. It appears that Sparse Classification out-performs KNN, while state-labeled decoding provides better per-formance than word-labeled decoding. In all conditions the per-formance increases with the size of the collection. However, the optimal window length is 10 frames for state-labeled decoding, but 35 frames for word-labeled decoding. 1.