Key information extraction from Broadcast in traffic domain

DOI

10.2991/aiie-15.2015.64How to use a DOI?

Keywords

automatic speech recognition, artificial intelligence, natural language processing, HMM, CRF

Abstract

With the development of Artificial Intelligence, Automatic Speech Recognition has become an important field of cognitive computing in terms of intelligent vehicle. This paper introduces the application of Speech Recognition and Natural Language Processing on intelligent vehicle that extracts key information from traffic broadcast. This system consists of three modules, speech detection module, keywords output module, and structured information extraction module. The speech detection module uses Non-negative matrix factorization to make speech enhancement, which separates the speech from the background music. Using Mel-frequency Cepstral Coefficients (MFCCs) extracted from the speech as features, it trains the model using Hidden Markov Model (HMM). Language model like n-gram improves the performance of the recognizer. In structured information extraction module, Conditional random field (CRF) is used to make partial parsing. Then it fills the slots, and outputs structured traffic condition information. We give the experiment results of every module.

Copyright

© 2015, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - J.B. Gu
AU  - Y.F. Xie
AU  - Y.Y. Wang
AU  - X. Xie
PY  - 2015/07
DA  - 2015/07
TI  - Key information extraction from Broadcast in traffic domain
BT  - Proceedings of the 2015 International Conference on Artificial Intelligence and Industrial Engineering
PB  - Atlantis Press
SP  - 229
EP  - 232
SN  - 1951-6851
UR  - https://doi.org/10.2991/aiie-15.2015.64
DO  - 10.2991/aiie-15.2015.64
ID  - Gu2015/07
ER  -