Speech technology goes mobile

Despite the recent disintegration of premier speech technology company, Lernout & Hauspie Speech Products NV (L&H), the technology that it helped pioneer continues to move forward off the desktop PC – which put L&H on the map – and onto mobile devices.

The problems at Ieper, Belgium-based L&H multiplied this week when former chief executive officer Gaston Bastiaens was arrested at his home in Massachusetts, nearly one month after co-founders Jo Lernout and Pol Hauspie were arrested in Belgium. L&H’s U.S. headquarters are in Burlington, Mass.

The executives were arrested for questioning on charges of financial fraud, stock price manipulation, and other charges, authorities said.

Whereas leading companies such as Dragon Systems Inc. and Dictaphone Corp. are already gone, the fate of L&H is still up in the air, a number of industry analysts have said. The Belgian government is expected to rule next week on whether or not L&H can sell off its assets as separate companies rather than as individual assets. All media contact has been suspended until after the Belgian courts rule.

If L&H is split up, the plan is to divide it into four separate companies each seeking its own buyers, said Bill Meisel, a speech technology analyst and president of TMA Associates in Tarzana, Calif., who was recently briefed by L&H.

The four companies would include its health care operations, formerly part of Dictaphone, and a call center and recording business that holds a major market share in recording 911 calls, also from Dictaphone, Meisel said.

L&H’s Mendez Translation services, consisting mainly of human translators, would also be sold as a group. The same might be true for the Speech and Language Systems division, which might keep the name L&H, Miesel said.

“It might make sense for Microsoft to look at their Language Systems division,” Meisel added.

Although L&H may soon disappear, other companies are quickly advancing with a different perspective on speech technology.

Called multimodal access, companies like Auvo Technologies in Itasca, Illinois, are targeting speech technology toward the increasing need to offer mobile solutions to their work force.

Multimodal technology uses a combination of graphical, touchscreen, and voice interfaces to gain access to data.

Microsoft, which launched Office XP this week with speech recognition built in, is also developing multimodal access technology for its Win CE handheld platform and the forthcoming Windows XP.

“We are working on this project. You click with the stylus to select a field then speak to fill in the field,” said Alex Acero, manager of the speech research group at Microsoft in Redmond, Wash.

Auvo, using the speech engine from Nuance Communications, is currently selling to wireless carriers and is in a trial agreement with Vodaphone AirTel in Madrid, Spain. The company will officially launch its multimodal handset browser in the third quarter.

The technology has two parts: a client speech-compression component for transmission over an IP network, and a back-end platform on the network side for converting the speech to data.

In the United States, multimodal handsets will have to wait. The current cellular technology cannot handle a voice and data channel simultaneously as can the GSM (Global System for Mobile communications) wireless network in Europe.

The likelihood of the alleged misdeeds of L&H executives tarnishing the image and thus setting back the deployment of speech technology is small, most industry experts agreed.

“Companies in every industry have had problems including businesses where credibility is critical, such as financial services. If a bank is dishonest [it] doesn’t mean we stop banking,” said Amy Wohl, a speech technology analyst and editor of Opinions, an online newsletter in Narbuth, Penn.