DocumentCode
2920648
Title
Mathifier — Speech recognition of math equations
Author
Batlouni, Salim N. ; Karaki, Hala S. ; Zaraket, Fadi A. ; Karameh, Fadi N.
Author_Institution
American Univ. of Beirut, Beirut, Lebanon
fYear
2011
fDate
11-14 Dec. 2011
Firstpage
301
Lastpage
304
Abstract
Speech recognition has become widely used across many applications. Telephone systems can route a phone call based on what the caller says, control systems can respond to actions said by the controller, and mobile phones can recognize the speech of a contact´s name and call the respective contact directly. However, speech recognition has found little use in recognition of textual material due to the large dictionary and hence large word error rates. Mathifier constricts the speech recognition to math equations; it takes as input math formulas presented in the form of user speech and produces the equations in digital mathematical form. The smaller dictionary and the specific grammar structure of the math equations help restrict the problem of the recognition process. The program has room for smartly guessing words based on the grammar structure and thus resulting in a lower error rate and better recognition. Mathifier uses Sphinx, a modular speech recognition tool from CMU, and adapts it to recognize math equations and convert them into latex form in real time.
Keywords
mobile handsets; speech recognition; CMU; Sphinx; control systems; math equations; mathifier; mobile phones; phone call; speech recognition; telephone systems; Equations; Grammar; Hidden Markov models; Java; Mathematical model; Speech; Speech recognition;
fLanguage
English
Publisher
ieee
Conference_Titel
Electronics, Circuits and Systems (ICECS), 2011 18th IEEE International Conference on
Conference_Location
Beirut
Print_ISBN
978-1-4577-1845-8
Electronic_ISBN
978-1-4577-1844-1
Type
conf
DOI
10.1109/ICECS.2011.6122273
Filename
6122273
Link To Document