Alexandre Trilla, PhD - Research Engineer | home publications
 

Blog

-- Thoughts on data analysis, software development and innovation management. Comments are welcome


Post 8

ASR in the Asus EeePC

26-Dec-2008

Having fun with the present that Santa Claus brought for me, the UMPC Asus EeePC 900, I am now enoying the device bringing it to one of my fields of study, Automatic Speech Recognition (ASR). The computer has an embedded speech recognition engine that is used to recognize the user's voice commands to launch some applications. It has a modus operandi similar to Magnus, my Master's Thesis: the input speech is analysed continuously in order to determine the spoken commands.

As a difference, the EeePC speech recognizer requires a first keyword to start the recognition process. No previous training is required by the user. The application loads fast, which impedes heavy acoustic models to be loaded on the fly. Then, could it be a Dynamic Time Warping (DTW) ASR engine? This would explain its somewhat poor performance (the system gets confused from time to time) since the system is not adapted to the user's voice. This was one of Dr. Iriondo's suggestions at my Master's Thesis defense in order to improve the performance of Magnus, which uses the acoustic models produced by the LDC speech samples, thus not adapted to the Catalan phonetic set. Since the command set is not big, DTW could yield nice results.

The only useful information I've found about this application is its process name, SPAVoiceCommand, and Google has been useless since no result has been returned for its query.



All contents © Alexandre Trilla 2008-2025