Analysis of keyword spotting performance across IARPA babel languages
William Hartmann, Damianos G. Karakos, Roger Hsiao, Le Zhang, Tanel Alumäe, Stavros Tsakalidis, Richard M. Schwartz
Abstract
With the completion of the IARPA Babel program, it is possible to systematically analyze the performance of speech recognition systems across a wide variety of languages. We select 16 languages from the dataset and compare performance using a deep neural network-based acoustic model. The focus is on keyword spotting using the actual term-weighted value (ATWV) metric. We demonstrate that ATWV is keyword dependent, and that this must be accounted for in any cross-language analysis. Further, we show that while performance across languages does not track with any particular feature of the language, it is correlated with inter-annotator agreement.
BibTeX
@inproceedings{icassp2017_analysisofkeywor,
title = {Analysis of keyword spotting performance across IARPA babel languages},
author = {William Hartmann and Damianos G. Karakos and Roger Hsiao and Le Zhang and Tanel Alumäe and Stavros Tsakalidis and Richard M. Schwartz},
booktitle = {ICASSP 2017},
year = {2017}
}