SPEECH TECHNOLOGY AS THE MOST NATURAL MAN / MACHINE INTERFACE TO ACCESS INFORMATION.

 Jo Lernout
 President
 Lernout & Hauspie Speech Products
 Sint Krispijnstraat 7
 8900 Ieper Belgium
 Tel: +32-57-228888

ABSTRACT

The availability and quality of information that one can access today, be it voice, video, text or data, is constantly improving and expanding. Furthermore, with the growing infrastructure of networks and decreasing costs of this information "on-line", users are taking advantage of this information be it from the office, from home or while traveling abroad. However, the shortcomings in today's user-interface to this information jeopardizes the actual amount of knowledge gained by the user.

Speech technology (speech recognition and text-to-speech) not only increases the efficiency of the user, it also enables the telephone to be used as a tool (as is the computer) to facilitate access to this information. This is attractive to the user as he or she has access to the information wherever they are. This is attractive to the multimedia software developer because a speech-interface creates the possibility to both "humanize" the man-machine interface of their product line as well as add new functionality and features to their product offering. This is attractive to the service provider as it opens up whole new markets to the services they provide.

Speech recognition (the ability of electronic equipment to recognize human speech as a form of input), and text to speech (to transform computer readable text into speech), will be an integral part of the man-machine interface, in combination with all other means of interfacing.

The purpose of this discussion is not to go into technical details of how these technologies work, but to examine;

 (1) why, only now, are we turning to speech technology;
 (2) the current status of speech recognition and text-to-speech and what
     is coming in the future;  
 (3) in what applications (information access) are the technologies being
     applied today.

Moreover, we will examine how the synergy of these two speech technologies (speech recognition & text-to-speech) enables users to have a dialog with an application. Natural language understanding, continuous speech recognition and natural sounding text-to-speech in many languages are essential ingredients in creating a natural man-machine interface to access information.


BIOGRAPHY

In 1987, Jo Lernout started Lernout & Hauspie Speech Products, together with Pol Hauspie. Prior to Lernout & Hauspie, Mr. Lernout held sales positions at Merck Sharp & Dohme (Belgium) and Bull Computers. Mr. Lernout was also the marketing director at Wang (Belgium), and the general manager at Barco Industries Inc. in the US. Mr. Lernout's company has experience in the development and marketing of speech and music compression, multilingual text-to-speech and speaker independent speech recognition. Mr. Lernout has been a guest lecturer at several universities, world economic meetings and several voice conferences in Europe, Japan, South Africa, and US.

Back to Conference Programme Page