EVALUATION OF ROBUSTNESS IN ASR FOR DIFFERENT ‘FRONT-END’ METHODS

  • 1 Faculty of Mathematical and Physical Engineering – Polytechnic University of Tirana

Abstract

Some feature extraction methods suffer performance degradation in different environments. So it has become a necessity to search for new methods that perform better in different types of conditions. Therefore we can make a comparison of the new found methods to evaluate their performance and to determine which is best in multi-condition tests in order to have e more robust ASR system.

Keywords

Article full text

Download PDF