Home // ICDS 2015, The Ninth International Conference on Digital Society // View article

Improving ASR Recognized Speech Output for Effective Natural Language Processing

Authors:
Chandrasekhar Anantaram
Sunil Kumar Kopparapu
Nikhil Kini
Chiragkumar Patel

Keywords: evolutionary development; artificial development; speech recognition; natural language processing.

Abstract:
The process of converting human spoken speech into text is performed by an Automatic Speech Recognition (ASR) system. While functional examples of speech recognition can be seen in day-to-day use, most of these work under constraints of a limited domain, and/or use of additional cues to enhance the speech-to-text conversion process. However, for natural language spoken speech, the typical recognition accuracy achievable even for state-of-the-art speech recognition systems have been observed to be about 50 to 60% in real-world environments. The recognition is worse if we consider factors such as environmental noise, variations in accent, poor ability to express on the part of the user, or inadequate resources to build recognition systems. Natural language processing of such erroneously and partially recognized text becomes rather problematic. It is thus important to improve the accuracy of the recognized text. We present a mechanism based on evolutionary development to help improve the overall content accuracy of an ASR text for a domain. Our approach considers an erroneous sentence as a zygote and grows it through an artificial development approach, with evolution and development of the partial gene present in the input sentence with respect to the genotypes in the domain. Once the genotypes are identified, we grow them into phenotypes that fill the missing gaps and replace erroneous words with appropriate domain words in the sentence. In this paper, we describe our novel evolutionary development approach to repair an erroneous ASR text to make it accurate for further deeper natural language processing.

Pages: 17 to 21

Copyright: Copyright (c) IARIA, 2015

Publication date: February 22, 2015

Published in: conference

ISSN: 2308-3956

ISBN: 978-1-61208-381-0

Location: Lisbon, Portugal

Dates: from February 22, 2015 to February 27, 2015