15th February 1999 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
VOCALIS IN STRONG POSITION AS SPEECHDAT II GENERATES DATA FOR EXTENDING SPEECH RECOGNITION TO ALL MAJOR EUROPEAN LANGUAGESThe recent successful completion of a major pan-European research project has resulted in one of the largest, most wide-ranging databases of spoken language yet produced. The project, SpeechDat II, has involved nearly 20 of Europe's leading technology companies, academic centres and research institutes. As a leading participant, Vocalis, the Cambridge-based speech recognition telephony company, now has access to all the data generated, which will be of great value in its development of future advanced speech recognition (ASR) systems. Samples of language spoken in realistic situations are vital for the development of accurate ASR, but until now the databases available have been predominantly English. A central aim of the SpeechDat II project was to counter this bias. Andy Walker, Product Marketing Manager for Vocalis explains, "The spoken language resources produced by SpeechDat II, such as speech databases, pronunciation lexica and related tools, will enhance the availability and quality of a wide range of voice-driven telephone services. The production of these resources for an extensive range of European languages, including minority languages and regional dialects, will allow users to access these services in their own mother tongue, and means continental companies will be able to compete more effectively." The data collected by SpeechDat II covers all 11 official European languages, as well as dialects, such as the Swiss variants of French and German. Three types of speech samples have been gathered: from fixed and mobile telephone networks for training and testing recognisers, and a third for developing speech verification applications. This has resulted in the production of a total of 28 databases. "A further important benefit of the SpeechDat II project is that all these resources have been produced to a common standard and in a format which will support re-use across a wide range of applications," Walker says. "This will help to reduce the cost of creating new services." SpeechDat II has also developed a set of best practice guidelines and quality assurance practices for producing speech resources. The data it has produced will also be extremely valuable for future research into speech recognition. Further information The complete list of partners in SpeechDat II is as follows:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Back to Top | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
About Vocalis Vocalis works with organisations to create contact centre solutions that build the organisations business and brands. These solutions can turn your contact centre from a cost to a profit centre by ensuring it builds your brand, helps your business be more competitive, and increases customer loyalty through more effective and efficient service. Vocalis was formed in 1993, is publicly traded on FTSE Stock Market and has been listed since July 1996. Vocalis brings back the reassurance of the most personal human touch in business - putting voice to work. The Vocalis Website is at Email: [email protected] |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
For further Vocalis Group information, contact:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||