At
Seventh Sense Software we are actively developing
state-of-the-art algorithms and visualisation tools for use in large
vocabulary conversational speech recognition.
We are also pioneering the development of Unicode-compliant state-of-the-art multilingual speech recognition systems. A software prototype for recognition of isolated Bengali words and phrases has been implemented and was successfully demonstrated at the Secretariat of the Ministry of Science and Technology in August 2003 before senior government officials and members of the press. A screenshot of the prototype system is viewable here.
By freeing itself from the constraints of the textbook and to avoid the well-known drawbacks of standard algorithms, Seventh Sense Software has undertaken the development of a toolbox of novel and highly sophisticated techniques for the processing of raw speech data. These promise to break through the technological barriers which have to date stifled progress in development of high accuracy, versatile, speech recognition systems.
Work is ongoing with the aim of developing the world's first commercial quality large vocabulary continuous speech recognition system for the Bengali language.
[Course announcement:
C++ Programming for Scientists
and Engineers]
[Discussion group: http://groups.yahoo.com/group/seventh_sense]