Home > Articles

  • Print
  • + Share This
This chapter is from the book

Voice Recognition—So Much Talk

You may have seen the commercial in which a driver is approaching an intersection with a red light and he says, "Green light" and the light immediately changes to green. Neat stuff, if it actually worked under real-life conditions.

This technology is known as voice recognition or speech recognition. It is the natural interface for wireless devices and one of the more obvious methods of easing concerns about drivers who become distracted while trying to punch in a number in those tiny keypads while using their cellphones. You can simply speak the name or number you want to call, either into a handheld phone or a well-placed microphone dedicated to this purpose.

Most of the time.

Several companies around the world have spent millions trying for at least 25 years to get this technology to work, but it still has some serious bugs. For one thing, it's often not very accurate; that is to say, the technology does not accurately recognize exactly what you're saying and respond accordingly. Another problem is ambient noise. Try using one of these systems at a noisy trade show, at an airport gate during a public address announcement, or in your car with the radio on.

Commercial products and services have been available for some time in some specific, well-controlled applications. Getting airline flight information is one that seems to work most of the time. Voice-activated consumer products are also available, but are technically limited.

Voice portal services such as BeVocal, ShopTalk, and Tellme Networks are gaining in popularity. Sprint PCS now offers its customers a service called Voice Command, which enables users to create a voice-accessible address book. Yahoo! and Lycos have introduced a suite of speech tools and services that give consumers access to their content by telephone. Lernout & Hauspie, a leading speech technology company until it entered bankruptcy proceedings and then closed up shop, had announced plans to enter the wireless communications market with a system that lets mobile phone users access information on the Web, such as traffic reports and movie listings. PDA maker Palm has teamed with SpeechWorks International to add speech recognition to Palm's Web-based calendar service. Motorola has introduced its iRadio Internet system for automobiles with Internet access, a directory dialer, and address book, and the ability to send and receive e-mail through its voice recognition feature. All of these efforts should lead to improved voice recognition services.

Like so many other technologies, the Internet will be the driving force behind getting voice access into the network. Increasingly, the technology, because it is so easy to use, actually presents carriers and other wireless service providers with an attractive alternative to existing industry technology standards such as the Wireless Application Protocol (WAP). As an alternative to WAP, speech-recognition enables users to access Internet content hands-free.

One of the first things that has to happen to make voice recognition work for everyone is the creation of a technical standard, and that process is well underway. Version 1.0 of the VoiceXML (Voice eXtensible Markup Language) specification has been accepted as a standard by the World Wide Web Consortium (W3C). The W3C's Voice Browser Working Group has agreed to base its efforts to develop a standard on VoiceXML. This is a spec that could provide a high-level programming interface to speech and telephony resources for application developers, service providers, and equipment manufacturers.

Unfortunately, more than two years after Motorola, IBM, Lucent Technologies, and AT&T helped form the VoiceXML Forum to bring technical standards to voice recognition, there is still no way of ensuring that any of these systems can talk to each other.

Interoperability concerns have begun to slip through the standards development cracks and have taken on more of a competitive marketing track, with voice recognition companies introducing their own "open standard" systems and selling their products as modules that can be updated or changed-out as new accessories and technologies are developed. The result is that few of these products may work together, or they won't work together very well.

Wireless service providers can't wait for these developments to kick in. They're convinced that easy access to different services through voice recognition will increase their traffic—and their revenues. Automated speech-recognition-enabled services could also produce cost savings for the carriers; the cost of processing a phone call using an automated directory assistant is about one-tenth the cost of processing the same call using operator assistance. Bottom line, look for voice portal companies to introduce more sophisticated and useful applications. In fact, voice access to e-mail and Web-based information and services appears to be well on its way to becoming a primary consumer interface for a variety of portable electronic products, including electronic games.

  • + Share This
  • 🔖 Save To Your Account