NEW YORK-Voice Signal Technologies Inc. will begin commercial sale later this summer of its VoiceTag wireless communications speech- recognition solution, designed to be embedded in the device instead of relying on computer servers or distributed architecture.
The privately held company, headquartered in Woburn, Mass., is involved in discussions with manufacturers of wireless handsets and handheld computers, said Stewart Sims, executive vice president of marketing.
Incorporating voice recognition in wireless devices confers several advantages, including availability even in the absence of a network connection and responsiveness unaffected by delays in network transmissions, the company said.
VoiceTag’s said its embedded speech interface software can be implemented now on existing mobile platforms without the need for new telecommunications networks, front-end processing standards or other technological advances.
In addition to voice-activated dialing, VoiceTag provides command-and-control capabilities to allow device users to sort through all the functions on their mobile communicators by means of single, spoken commands. Instead of searching through on-screen icons, users can activate the Internet browser by saying “browser.” They also can view the records of their most recent incoming calls by saying “calls received.”
“There are so many features on today’s mobile phones that consumers have a hard time finding them all. By adding speaker-independent command and control to mobile phones, we are enabling users to access features directly, with intuitive spoken words,” Sims said.
The voice-recognition software is designed to adapt to the speech patterns and accents of individual users. It also offers “continuous digit recognition,” a phrase that describes its ability to let end users tell their phones to call a given number.
“Today’s mobile-phone users will increasingly have to rely on hands-free dialing as a safe and convenient way to use their mobile phone in the car. Continuous digit recognition, one of the hardest technical problems in speech recognition, substantially enhances that capability,” Sims said.
The speech-recognition developer sees significant opportunities for VoiceTag in the new telematics applications that automobile manufacturers are developing. The software’s platform independence also means it can be adapted readily to handheld computers.
Because Voice Signal understands that speech technology “is a very powerful tool but not necessarily the total solution for interface problems,” it has designed its software “to support an intelligent, task-oriented, multi-modal user interface,” the company said. This means VoiceTag can work with other interfaces that rely on vision, pointing and button pushes.
“An intelligent interface does not seek to prove that one interface technology is superior to others; rather, it seeks to make use of the most suitable technology or combination of technologies for the specific tasks,” Voice Signal said.
“This means that focusing strictly on measurements such as recognition accuracy and noise robustness as the criteria for a successful speech- enabled interface is too restrictive. Voice Signal believes the true measure of a successful interface is the time it takes to complete the whole task.”
To provide response times as fast as 20 milliseconds, VoiceTag employs techniques that include continuous listening, even while processing, and proprietary noise cancellation capabilities.
Established in 1995, Voice Signal Technology has a research team comprised of scientists and engineers formerly affiliated with companies including Siemens Speech Research and Siemens Electronics, Nokia Corp., Texas Instruments, Lucent Technologies Inc. and Analog Devices. VoiceTag is currently available on many common mobile-phone platforms, including those made by Texas Instruments and Analog Devices.
