International Journal of Computer Applications |
Foundation of Computer Science (FCS), NY, USA |
Volume 63 - Number 21 |
Year of Publication: 2013 |
Authors: Himanshu Rai Goyal, Shashidhar Koolagudi |
10.5120/10589-5429 |
Himanshu Rai Goyal, Shashidhar Koolagudi . Hindi Number Recognition using GMM. International Journal of Computer Applications. 63, 21 ( February 2013), 25-30. DOI=10.5120/10589-5429
This paper aims at designing and implementation of Hindi number recognition system using the microphone and mobile recorded speech. Spectral features known to represent phonetic information are used as the features to characterize different Hindi digits. Gaussian mixture models (GMM) are used to develop the digit recognition system. This paper focuses on the ten basic Hindi digits where '0' is pronounced as 'shunya' to '9' is pronounced as 'no'. Data has been collected separately from male, female and child speakers using microphone and mobile phone device. The experimental results show that the overall accuracy of digit recognition is 98. 9\% in the case of microphone recorded speech and 96. 4\% in the case of mobile phone recorded speech.