TY - GEN
T1 - Part of speech tagging approach to designing compound words for arabic continuous speech recognition systems
AU - AbuZeina, Dia
AU - Elshafei, Moustafa
AU - Al-Khatib, Wasfi
PY - 2011
Y1 - 2011
N2 - Misrecognition of small words is one of the factors that lead to suboptimal performance in automatic continuous speech recognition systems. In general, errors generated from small words are much more than errors in long words. Therefore, compounding some words (small or long) to produce longer words is welcome by speech recognition decoders. In this paper, we present a novel approach to artificially generate compound words using part of speech tagging. For this purpose, we consider two Arabic pronunciation cases that usually occur together without any silence: a noun followed by an adjective, and a preposition followed by any other word. To collect the candidate compound words, we use Stanford Arabic tagger to tag all words in our Baseline transcription corpus. Using Sphinx 3, we test the proposed method on a 5.4 hours speech corpus of modern standard Arabic. The results show significant improvement, with the word error rate being reduced by 2.39%.
AB - Misrecognition of small words is one of the factors that lead to suboptimal performance in automatic continuous speech recognition systems. In general, errors generated from small words are much more than errors in long words. Therefore, compounding some words (small or long) to produce longer words is welcome by speech recognition decoders. In this paper, we present a novel approach to artificially generate compound words using part of speech tagging. For this purpose, we consider two Arabic pronunciation cases that usually occur together without any silence: a noun followed by an adjective, and a preposition followed by any other word. To collect the candidate compound words, we use Stanford Arabic tagger to tag all words in our Baseline transcription corpus. Using Sphinx 3, we test the proposed method on a 5.4 hours speech corpus of modern standard Arabic. The results show significant improvement, with the word error rate being reduced by 2.39%.
KW - Modern Standard Arabic
KW - compound words
KW - language model
KW - part of speech tagging
KW - speech recognition
UR - http://www.scopus.com/inward/record.url?scp=82955213595&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-25483-3_27
DO - 10.1007/978-3-642-25483-3_27
M3 - Conference contribution
AN - SCOPUS:82955213595
SN - 9783642254826
T3 - Communications in Computer and Information Science
SP - 330
EP - 338
BT - Informatics Engineering and Information Science - International Conference, ICIEIS 2011, Proceedings
ER -