Publication Details
Progress in the BBN Keyword Search System for the DARPA RATS Program
Hsiao Roger (FIT)
Zhang Le (FIT)
Karakos Damianos (FIT)
Mallidi Sri Harish (FIT)
Karafiát Martin, Ing., Ph.D. (DCGM)
Veselý Karel, Ing., Ph.D. (DCGM)
Szőke Igor, Ing., Ph.D. (DCGM)
Zhang Bing
Nguyen Long
Schwartz Richard (FIT)
speech recognition, KWS, MLP, DNN
This article is about the progress in the BBN Keyword Search System for the DARPA RATS Program (Robust Automatic Transcription of Speech).
This paper presents a set of techniques that we used to improve our keyword search system for the third phase of the DARPA RATS (Robust Automatic Transcription of Speech) program, which seeks to advance state of the art detection capabilities on audio from highly degraded radio communication channels. The results for both Levantine and Farsi, which are the two target languages for the keyword search (KWS) task, are reported. About 13% absolute reduction in word error rate (from 70.2% to 57.6%) is achieved by using acoustic features derived from stacked Multi-Layer Perceptrons (MLP) and Deep Neural Network (DNN) acoustic models. In addition to score normalization and score/system combination for keyword search, we showed that the false alarm rate at the target false reject rate (15%) was reduced by about 1% (from 5.39% to 4.45%) by reducing the deletion errors of the speech-to-text system.
@inproceedings{BUT111666,
author="Tim {Ng} and Roger {Hsiao} and Le {Zhang} and Damianos {Karakos} and Sri Harish {Mallidi} and Martin {Karafiát} and Karel {Veselý} and Igor {Szőke} and Bing {Zhang} and Long {Nguyen} and Richard {Schwartz}",
title="Progress in the BBN Keyword Search System for the DARPA RATS Program",
booktitle="Proceedings of Interspeech 2014",
year="2014",
pages="959--963",
publisher="International Speech Communication Association",
address="Singapore",
isbn="978-1-63439-435-2",
url="http://www.isca-speech.org/archive/interspeech_2014/i14_0959.html"
}