telecoms-bar
FT Telecoms May 16 2001 - Voice recognition
Speech technology could breathe new life into a universal communications tool
by Geoffrey Nairn
Published: May 14 2001 16:35GMT | Last Updated: May 16 2001 12:01GMT
image

Few consumers in developing regions can afford a PC and users in Asian countries are put off by the difficulty of typing complex characters using a PC keyboard designed for western languages.

These problems have led the telecommunications industry to look to speech technology to deliver interactive services in countries where PC penetration is low.

"The phone is the most ubiquitous terminal out there," says Minesh Patel, technical director at Vocalis, a UK-based speech technology vendor, which has customers in Malaysia, Brasil, South Africa and Mexico, as well as Europe.

Companies such as Vocalis hope the mobile phone boom will drive the uptake of speech technology around the world. While consumers in developing countries may still wait years for a fixed line, cellular phones are often easier to obtain and increasingly affordable.

South Africa, for example, now has 8m cellular subscribers - ten times more than the number of PC owners. The mobile phone was once reserved for a wealthy minority in South Africa but the introduction of pre-paid packages has caused an explosion in mobile phone ownership. By 2003, half South Africa's population could own one.

Strategy Analytics, a US research company, predicts 1.7bn users of mobile phones around the world by 2006. The main drivers of growth will not be North America and Europe - these regions will account for just 20 per cent of growth - but emerging markets such as Latin America.

Brazil is the most promising telecommunications market in Latin America. Since the privatisation of Telebras in 1998, foreign carriers have flocked to the country attracted by its 170m people and its underdeveloped telecommunications market.

Brazil hopes to have up to five mobile operators competing in each of its regions. The speech industry believes the intense competition will encourage operators to build voice portals to attract customers and generate additional revenues.

Avaya, the US equipment supplier, recently showed its voice portal technology at a trade fair in Sao Paolo. The portal enables rapid access to a wealth of resources, using a wireline or wireless terminal. For example, a user caught in a traffic jam can listen to e-mail messages and respond using his or her own voice.

Users can also get a list of local restaurants and make a booking, or, this being Brazil, listen to the previous day's telenovela.

Avaya's voice portal is based on speech recognition technology from Nuance, the US company. Banco Bradesco, which claims to be Brazil's and South America's largest private bank, is already using Nuance technology.

It has created a voice-driven bill payment system, called FoneFacil, which allows customers to electronically pay bills by speaking to the system in the same way that they would speak to a live person.

At the beginning of the call, they read their account number and then read the digits from a 48-digit code that appears on every bill in Brazil. After that, the system reads back the details of the bill, asks for confirmation and arranges to pay the bill.

SpeechWorks, the main competitor to Nuance, also sees a promising market for speech technology in Latin America but, according to Steve Chambers, its vice president of worldwide marketing, the international market with greatest potential is currently the Asia Pacific region.

"In Asia, the number of new phone subscribers is 12 to 20 times the number of new web subscribers," he says. Nuance is also betting heavily on the Asian market and earlier this month it announced a big expansion in Asia with a new regional headquarters in Hong Kong and an office in Seoul.

China has long been seen as a promising market for speech technology. The country has 24m internet users and 131m phone lines but, relative to the total population, there is only one phone line for every 10 people. Mobile phones could boost that ratio dramatically in coming years and there is keen interest in developing speech recognition systems to bring voice-enabled interactive services to China.

In the past two years, IBM, Microsoft and Intel have all invested in speech recognition research initiatives in China. The specialist speech companies have also jumped on the Asian bandwagon.

Nuance's software is available in 22 languages, including Cantonese, Mandarin, Singaporean English and Japanese. Philips Speech Processing, a European rival, claims it developed the world's first Mandarin language voice portal for KG Telecom, a Taiwanese telephone company.

In emerging markets, western vendors have the benefit of a portfolio of established customers who have been using speech technology for several years. But the Asian market is a difficult one for western vendors and not just because of the technical difficulties in adapting their speech technology to handle Asian languages.

"Everyone asks us about the technology but people do not pay as much attention to the cultural issues in Asian countries," says Mr Chambers.

There is also growing local competition in the Asian speech market. Hong Kong-based start-up Weniwen - it means "ask away" in Mandarin - is developing technology to enable Chinese users to access the internet using voice commands via a mobile phone.

Weniwen has attracted investors from Asia and Silicon Valley, including Mark Hoffman, founder of Commerce One, the US e-commerce software house. Another Hong Kong speech venture is iSilk, which recently won $10m of funding from a group of investors including Hutchison Whampoa, Ericsson and Investor.

InfoTalk, a third Hong Kong start-up, claims to have developed the world's first tri-lingual speech recognition solution, which is used for reading gas meters, as well as a system for real-time stock quotes that can handle Chinese.

At the height of the dotcom boom, China spawned many dubious internet ventures as investors fixed their eyes on a market with a potential 1.3bn consumers but forgot that the majority have never used a PC keyboard.

Now that some sanity has returned, it is clear that China and other developing internet markets are not going to blossom overnight. Nevertheless, speech technology could hasten the day when the internet becomes a truly global phenomenon that reflects the world's diversity of cultures and languages.