Recognizing and Investigating Spoken Arabic Alphabet

Author

Alotaibi, YousefAjami

Author_Institution

King Saud Univ., Riyadh

fYear

2007

fDate

15-18 Dec. 2007

Firstpage

171

Lastpage

175

Abstract

Alphabet recognition is needed in many applications for retrieving information associated with the spelling of a name, such as telephone, addresses, etc. In this paper Arabic alphabets were investigated from the speech recognition problem point of view. The system was an isolated whole word speech recognizer. Spoken Arabic alphabet has more than one set, each of which contains members of alphabets that are acoustically very similar. This recognition system achieved 83.34% correct for alphabets. The spoken alphabet "Hamzah" got almost 100% recognition rate, but the worst performance was encountered with alphabet "Baa", "Taa", "Thaa", "H_aa", "Thaal", "Raa", "Seen", "T_aa", "Dhaa", and "Faa".

Keywords

hidden Markov models; natural language processing; speech recognition; HMM; hidden Markov models; information retrieval; speech recognition; spoken Arabic alphabet recognition; Application software; Educational institutions; Hidden Markov models; Information retrieval; Information technology; Mel frequency cepstral coefficient; Natural languages; Signal processing; Speech recognition; Testing; Alphabet; Arabic language; Baa-set; HMM; Speech recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Processing and Information Technology, 2007 IEEE International Symposium on

Conference_Location

Giza

Print_ISBN

978-1-4244-1834-3

Electronic_ISBN

978-1-4244-1835-0

Type

conf

DOI

10.1109/ISSPIT.2007.4458088

Filename

4458088