Other translations of this page: Eesti

Research and development of methods for Estonian speech recognition

Introduction

The project is concerned with the research, development and testing of methods for Estonian speech recognition and the implementation of speech recognition prototype systems for the Estonian language.

Main tasks of the project:

  1. determining optimal basic lexical units for Estonian LVCSR (such as syllables, (pseudo-)morphemes, data-driven units);
  2. development of statistical language modeling techniques using the determined lexical units; integrating long-term semantic dependencies into the language model;
  3. applying of acoustic model adaptation techniques;
  4. investigating speech recognition over mobile phone channels and/or in mobile devices;
  5. delivering optimal solutions for development of medium-vocabulary speech recognition systems;
  6. development of methods and algorithms for large/unlimited vocabulary speech recognition systems;
  7. implementation of speech recognition prototype systems.

For more information, please see the page in Estonian.

 
projects/tuvastus/tuvastus.en.txt · Last modified: 2010/09/23 09:36 by tanel