Application of Amharic Speech Recognition System to Command and Control Computer: an Experiment with Microsoft Word
No Thumbnail Available
Date
2003-07
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
This study explored the possibility of developing Amharic speech input interface to command
and control Microsoft Word. Towards this end, literature were reviewed on speech recognition,
application of speech recognition, HMM and its application in speech recognition, Amharic
speech recognition, HTK and human-computer interaction.
Speech input interface requires speech recognition system. To develop and test the required
Amharic speech recognition system speech data were recorded from 26 people (10 female and 16
male) in the age range of 20 to 35. 76.9% of the recorded data were used to train the recognizers
and the remaining data were used for testing the performance of recognizers. Two (fixed variance
and variable variance based models) HMM-based, speaker independent, small vocabulary,
isolated Amharic word recognizers were developed. The performance of these recognizers was
tested using the test data. Although both of them recognized all the test data correctly, the
performance of recognizer with variable variance performed better than the recognizer with fixed
variance in live environment. Thus, the recognizer with variable variance was further considered
for the development of the prototype Amharic speech input interface system.
Speech input interface requires communication interface that sends the recognized command
word to the application, Microsoft Word in this case. In this study the communication interface
was written using Visual Basic 6. To test the performance of the system as a whole, 18 randomly
selected command words were given to 6 people (3 command words for each) and these people
were asked to command Microsoft Word orally. The system performed 16 commands accurately
and only two command words were wrongly recognized and thus Microsoft Word performed
wrong actions. Finally, the prototype speech input interface system developed in this experiment
was integrated with Microsoft Word as a Macro for regular use.
The result of the experiment showed the feasibility of developing Amharic speech input interface.
Based on the result of the study and what has been learned in the course of the research
recommendations for further study were forwarded.
Description
Keywords
Speech Recognition