Andrew Cooper, Public Relations Executive, Ectaco

Q In a recent white paper from Ectaco, you mentioned that the "automated call centers" were costly and "carved up". Could you explain your thoughts on this notion?

A The considerably high cost of call centers is explained first of all by the fact that they are solutions for corporate clients. It would be difficult to imagine such a solution being implemented by an end user. We referred to the call-center market as being "carved up" because there are already major players in place, the consistent opportunism of which makes the market difficult to penetrate for newcomers, who in any case will not change the make-up of the market much. There is a limited segment, which by virtue of its limitations (independent of technical characteristics of call center such as recognition quality,) has a predictable sales volume significantly less than what analysts were expecting.

Q Also, in that same paper, you stated that speech technology companies needed to be in the right place at the right time to occupy a leading position in the anticipated market. What does this mean?

A There is an opinion expressed in the article that there is no mass market for speech technologies yet. In other words, there are many companies who are carrying out R&D for prospective solutions or creating speech systems made to order, but in order for the market to see significant growth, it needs a mass-market product. This product should have SR as a main functional capability, not a supplementary one. The company that first develops a consumer application and promotes it widely has a chance at becoming the market leader.

Q When do you think we will see the day that automatic translation over the telephone will become a reality?

A It is understandable that such a system is not possible without a powerful, universal system of speech recognition that assures a satisfactory result. There was a good quote on the latter from Gordon Moore, one of the founders of Intel and author of the well-known Gordon Moore's Law.

Talking at a briefing related to his receiving the Presidential Medal of Freedom, he commented, "Good speech recognition is the ability to talk to the computer as if were an individual. You can have an intelligent conversation with your computer. This would make computing available to people who are scared off by a keyboard. But I don't know if this is 10 years or 50 years away. I suspect it is closer to 50 years. The technology has been five to 10 years away for a long time".

Q What will it take for this to happen?

A Success in any market depends on a number of factors. For the speech technologies market, these can be divided into two groups: intellectual (improvement of algorithms, linguistic databases, hardware, etc.) and marketing factors. The latter seem to be the key. Any technology is created in order to help complete tasks for the user. A thorough understanding of customers' needs, correct positioning and a satisfactory price/quality ratio will go a long way in facilitating success.

Q What is unique about the Ectaco Universal Translator?

A Universal Translator is a mass-market device made to solve communication problems of people on trips abroad. The fact that it is in demand is demonstrated by the continued interest in the device from customers. Ever since its introduction on the market in Fall 2001, the UT has become Ectaco's best selling product.

Q Who could utilize your Universal Translator product?

A Anyone. The device is not only a good solution for travelers, but is also a good study aid for people learning a foreign language. Our surveys have shown that the UT has found use among all walks of life.

Q What is VoiceMethods and how is it different?

A VoiceMethods is first and foremost a powerful and universal set of tools for creating speech recognition systems. VoiceMethods technology is distinctive because of it small footprint and versatility, which allows it to be implemented in handheld devices and adapted for any hardware and software platform. Secondly, it is the name of an Ectaco-affiliated company, VoiceMethods LLC, which conducts speech research and development and marketing of its own solutions.

Q What are some of the obstacles you must overcome when putting embedding speech into mobile devices such as PDA's and phones?

A It seems that the path to creating such systems is strewn with technical and equipment limitations. The reality is that these technologies have reached a point to where quality speech recognition is possible on as small a device as a PDA. There are still a number of developmental nuances, however: algorithm improvement, the creation of quality linguistic databases. The majority of these issues have already been resolved, but the actualization of our know-how in other specific solutions besides the UT is a little way down the road, probably by about a couple of years.

Q In what vertical markets are you seeing the most demand for speech technology products and services?

A We would prefer not to divide the mass market in terms of vertical or horizontal markets. There is significant demand for our products in the U.S., and this demand is the basis for our economic success. Speech technology can be extremely versatile and its potential uses depend upon the richness of our imagination. Critical limitations to the development of such systems are not anticipated.

Q Tell us about your Machine Translation Toolkit and why it is important to your business?

A We realized long ago that in order to create a quality speech recognition system, certain elements of MT would have to be used. We have moved away from working on MT per se, but continue to improve our MT engine and use it in other solutions we develop.