IBM's Via Voice now learns Indian English

Mumbai: IBM SOFTWARE Group on Tuesday launched the Indian English version of speech recognition software Via Voice for the OEM (original equipment manufacturer) market. IBM Via Voice has been developed to suit the Indian English accent with the help of Speech and Software Technologies (I) Pvt. Ltd, a tata company.

In order to make the software recognise Indian accented speech, an Indian acoustic models with frequently used Indian words has been built, IBM said. speech recognition software is used for giving commands to computers, as a dictation tool and as a reader, among other things. Via Voice was developed using 40,000 sentences spoken by geographically distributed Indians. The starter set vocabulary consist of 62,000 words occuring frequently in the 16 million sentences collected from various Indian publications. Presently, the software has been taught the English launguare, and IBM does not have any plans to launch regional launguage versions.

The company did not quote a price for the product saying it would be sold directly to PC assemblers, manufacturers and the like. The software is pretty simple to use. After training it ot recognise and undestand a user's voice, the software can undestand most inflection of speech, making it less cumbersome than previous versions. Moreover, the software also sport a read out feature which can read out highlighted text. The software has been trained to recognise words such as lakh, crore, Mumbai, Chennai and other proper nouns which are unique to India. This is IBM first attempt to delivere a software product which has been specialy designed for the country.

Mr. Vishwesh Padmanabhan, VP, Tata IBM software Marketing, said: "The launch of the IBM'S strategy to bring in products suitable for the Indian market."


Source: The Economic Times, August 17, 1999, edition