The Hindu Business Line : Speech synthesis to improve man-machine interface

More flexibility
Speech recognition is a technology that uses natural speech as the interface between human beings and machines.
Plans to develop and roll out local language speech interface engine by March next.

Hyderabad , Dec. 18

Man-machine interface is set to get a lot better with speech synthesis technologies and new algorithms. They now give flexibility to capture different dialects than ever before.

From West and English-centric speech technology standards, there is an effort to encompass other languages. An Indian company has developed a local language interface engine in Telugu and expects to cover Hindi, Tamil, Gujarati and Marathi by March 2007.

The Managing Director of Bhrigus Software, Mr Nixon Patel, said after considerable debate it was felt that a purely West-centric model could not be considered a global standard. This is detrimental to a large non-English-speaking population.

Speech recognition

Speaking to Business Line on the current initiatives in speech technologies, Mr Patel said one of the challenges faced today is the limitation to recognise vocabulary and expand the scope of natural language interface.

Speech recognition is a technology that uses natural speech as the interface between human beings and machines such as computers, mobile phones and other devices. It converts speech into sets of complex algorithms based on a computer device or systems. Thereby bypassing the need to key in information.

The Speech Synthesis mark-up language (SSML) defines standards for speech synthesis. The World Wide Web Consortium has articulated the need to broaden the standards to cover other languages, as in the present form it is not truly international, Mr Patel said.

Other languages

For instance, people spend about 10 to 20 per cent of their precious time on an average in a day for travel in a car or other modes of transport. If this technology is used meaningfully, this precious time can be made use of in a better manner be it for official work or for entertainment.

Bhrigus has partnered with the International Institute of Information Technology, Hyderabad and researchers from the Carnegie Mellon University, who work under Prof. Raj Reddy, to develop a local language speech interface engine. They plan to roll this out in other languages by March next.

The range and scope of speech technologies and their applicability is multi-fold. Right from a basic interface of dialling a number on mobile phone, creating local language Web sites and interface with them, and the way people use PCs and other devices and automotives could change, he explained.

Patents

Bhrigus works with AT&T, Hewitt and Pfizer among others, providing speech enabled platforms. The company has filed for three patents and is now working closely with the World Wide Web consortium for new standards.

As a part of this standardisation initiative, an international meet is slated to be held in Hyderabad in February 2007 to deliberate on new standards, using extensions and additions for non-English language interface.