TECHNOLOGIES FOR INDIAN LANGUAGES: SOLVING REAL PROBLEMS OF REAL PEOPLE

Co-sponsored by: MEFGI

The talk covered the following technologies developed by MILE Lab for Tamil and Kannada: Text-to-speech conversion, printed text recognition, open vocabulary online handwriting recognition, Indic Keyboard IME for inputting most Indian languages on Windows, Linux and Android platforms and text extraction from camera captured scene images of display boards, road signs, etc. The technical challenges, the effort and enthusiasm necessary and systematic planning and data collection involved in developing these technologies were briefly covered. Some of these technologies were also briefly demonstrated. Finally, commitment of researchers towards developing realistic, multilingual speech recognition technology, called “Machine Listening” was covered. Machine listening is aimed at “listening with attention” to realistic human speech, involving more than one language (eg. Tamil and English), man-made non-speech sounds such as laugh, cough, humming, sneeze and throat clearing and environmental noise. Finally, the talk covered ongoing work on deep learning-based image super resolution and automated speech recognition for Tamil and
Sanskrit.

Speaker(s): Prof A G Ramakrishnan,

Agenda:

The agenda of the meeting was to discuss possible trends and work in the image processings, speech processings, and machine learning for indian languages.

Location:
Room: MC501, Seminar Hall
Bldg: Main Building
RAjkot Morbi Highway
Gauridad
Rajkot, Gujarat
360003