telecoms-bar
FT Telecoms May 16 2001 - Voice portals
Market is poised for take-off
by Neil McCartney
Published: May 14 2001 10:13GMT | Last Updated: July 18 2001 09:32GMT
image

The number of people accessing mobile internet services using voice portals could eventually match those using the keypads on their handsets, according to some analysts.

They argue that many mobile users would rather access internet services by speaking than by using a complicated process of pressing tiny buttons, and would often prefer to receive certain types of information by listening than by peering at small screens. And in some situations - for instance, while driving - there is a clear requirement for an access method that is hands-free and eyes-free.

Supporters believe that while the market for mobile voice portals has so far failed to take off, it is now finally set to do so. Frost & Sullivan, the market research and consultancy firm, predicts that the number of regular worldwide users will rise from only 200,000 at the start of this year to 203.8m at the end of 2007. It forecasts that the value of the market will rise from $3.6m to $5.06bn over the same period.

Co-author Carles Ferreiro, programme leader for mobile communications, says most of this will come from the additional voice traffic that will be generated by these services, although there will also be contributions from value-added services and advertising.

Omnitel of Italy was the first European operator to launch services in 1999. It has been followed by E-Plus of Germany and Telenor Mobile of Norway, but so far there are few others.

In the US, the most active company has probably been Sprint PCS, but again there are few commercial offerings. Most operators have been reluctant to launch voice portals because of concerns about the technology. Many have felt that the services on offer have been too limited to be attractive. More significantly, they have been worried about the number of errors that such systems tend to make in the noisy wireless environment.

Hakon Jacobsen, partner in communications and high technology at the Accenture consultancy, points out that many operators (particularly European ones) have been focused on other developments, including the introduction of Wireless application protocol (Wap), the technology that allows internet material to be displayed on the small screens of mobile handsets, and the implementation of high-speed third-generation (3G) networks. But this picture is now changing. Voice portals allow users to navigate internet menus and other applications by speaking voice commands into a telephone and to retrieve content either in the form of spoken responses or as text delivered in Wap or standard internet form.

Bernard Elliot, research director at Gartner Research, says that the performance of these voice interfaces has improved considerably in the last two years because of the advances made in speech-recognition (SR) and text-to-speech (TTS) technologies by companies such as Nuance, Lernout & Hauspie and SpeechWorks International.

Most services are menu-driven

Consumer portals are used by operators and independent companies to offer users services such as weather forecasts, driving directions, stock quotes and access to electronic messaging services. Enterprise voice portals are used by companies to provide staff and customers with specific information and services relating to their own businesses.

Most services are menu-driven: the customer is given a fairly limited range of options (such as a list of local information services which includes "restaurants"), and is asked to provide a spoken response, usually of one or two words. The system tries to match this reply to the sounds stored in its memory, and then moves on to the next stage (for instance, offering a range of restaurant types) before eventually giving the information required.

Such systems mark an advance on traditional interactive voice recognition approaches, where the customer enters responses via the keypad or speaks a number. But Traci Egli, another partner with Accenture's communications group, argues they still tend to be slow and cumbersome to use, and often have difficulty in recognising particular accents.

The next big step forward involves so-called natural language systems which can handle grammar. For the customer, these systems are much easier and faster to use because they can make requests such as "places to eat" or "where can I find a Chinese restaurant?".

However, such services appear to be a little way off. Chris Pearson, global m-business leader at consultants PricewaterhouseCoopers, says they require a considerable increase in processing power as well as advances in areas of social anthropology, such as how dialogue takes place. He adds that the greater the number of words in the system's vocabulary, the greater the possibility that it will misinterpret the customer's responses. This view is echoed by Simone Roberts of the Yankee Group consultancy. She says the first natural language portal is probably about two years away.

But progress is being made. The UK company Vox Generation last month launched a mobile voice portal platform which allows users to make commands with conversational English (from which it picks out key words) and does not need them to stick rigidly to a set menu of commands, because they can interrupt the service and change commands midstream.

Experts say such portals mark a significant step forward, because they offer a much more fluid experience, although they remain essentially menu-driven. And in March, Swedish mobile operator Europolitan Vodafone launched Europe's first voice portal based on technology using VoiceXML (VXML), the mark-up language for voice-user interfaces. VXML (see story on page four) allows internet content to be read directly by voice portal technology, without being converted first.

VXML is based on the World Wide Web Consortium's extensible mark-up language (XML) for the internet. It uses speech recognition and touchtone for input, and pre-recorded audio and text-to-speech synthesis for output.

Supporters claim its adoption will lead to a significant acceleration in the development of voice services.

Neil McCartney is editor-in-chief of Wireless Internet, www.mmwirelessinternet.com