Title :
Analysis of Urdu Syllables and Proposed Algorithm Structure for Determination of Speaker´s Speaking Rate
Author :
Mirza, S. ; Bin Liaqat, Hannan ; Hayat, S.A.
Author_Institution :
Dept. of Comput. Sci., COMSATS Inst. of Inf. Technol., Lahore
Abstract :
In this paper it is reported that definite relationships exist between the number of syllables in any Urdu text and the long vowels, short vowels and consonants these syllables contain. This information is useful in preparing an automatic algorithm to calculate speaking speed of a speaker. For every 100 syllables there are 38 short vowels, 58 long vowels and 3% of diphthongs. This percentage holds good for almost any number of syllables greater then about 50. The consonants have also been counted in syllables and found to be 135% of syllables. For every 100 syllables there are about 135 consonants, and this percentage holds true for syllables greater then about 50 syllables. These numbers help us calculate the probability of the long, short vowels and the consonants in any Urdu text, which can in turn be used in an algorithm to automatically calculate the speaking speed.
Keywords :
natural language processing; speaker recognition; Urdu syllables; Urdu text; speaker speaking rate; speaking speed; Algorithm design and analysis; Cepstral analysis; Information analysis; Information technology; Linear predictive coding; Mel frequency cepstral coefficient; Probability; Robustness; Speech;
Conference_Titel :
Multitopic Conference, 2007. INMIC 2007. IEEE International
Conference_Location :
Lahore
Print_ISBN :
978-1-4244-1552-6
Electronic_ISBN :
978-1-4244-1553-3
DOI :
10.1109/INMIC.2007.4557687