SignIt
Project that converts spoken Tamil to sign language
- Designed and implemented the first spoken Tamil to sign language conversion tools using concepts of Machine learning and phoneme rules for the language using the CMUSphinx API.
- Used tensorflow after a 90% hit rate to generate phonemes for Tamil words.
- Trained over data-set of 1500+ words and attained a Word Error Rate of ~21%
- Published āA Simple Tamil Speech Recognition System Based on CMUSphinxā in International Journal of Engineering and Advanced Technology (IJEAT), Vol.2 Issue. 3, 2019. ISSN: 2249-8958.
Publication: āA Simple Tamil Speech Recognition System Based on CMUSphinxā