ISCA Archive - Parcel sorting by speech recognition: a case study in vocabulary design (original) (raw)

Parcel sorting by speech recognition: a case study in vocabulary design

Clive R. Frankish, Dylan M. Jones

Two types of vocabulary were compared in a simulated parcel sorting task using automatic speech recognition. One consisted of place names in current use, the other of equivalent alphanumeric codes. When an isolated word recogniser was used, the rate of code entry was higher for the place name vocabulary. Although the proportion of correctly identified codes was similar for the two vocabularies, the greater redundancy of alphanumeric codes meant that more recognition failures could be detected automatically. These results are discussed in terms of their implications for the development of a practical system for parcel sorting.