Say What? DSR Goes Wireless

SpeechWorks International Inc. and Motorola have teamed up to develop what
they say is the first Distributed Speech Recognition prototype (DSR) deployed
over a wireless network.


Boston-based SpeechWorks , a provider of speech
recognition and text-to-speech technologies, said the prototype uses its
OpenSpeech Recognizer (OSR) speech recognition engine and DSR architecture.


Meanwhile, Armonk, N.Y.-based IBM launched WebSphere Voice
Server for Transcription, a speech recognition technology based on its
ViaVoice product that allows companies to implement speech-to-text
transcription solutions.


SpeechWorks and Schaumburg, Ill.-based Motorola have created
a field-force automation application that allows a sales representative to
check the status of any account by speaking into a handheld device from the
road, as well as a travel reservation application that allows travelers to
check the status of a flight while traveling.


The companies said the DSR architecture allows applications to combine local
speech processing that occurs entirely on the device with remote access to
network-based speech services.


Signal processing, including noise reduction, occurs on the device, which
then sends data over a digital network to a network-based speech service. The
network-based service processes the signal to determine the user’s request
and responds to the caller using a voice output, visual display, or both.


“Distributed speech recognition is essential to improving wireless speech
recognition especially in noisy environments like airports and train
stations,” said Chris White, director of speech products and multimodal in
Motorola’s Internet Content and Software Group.


“SpeechWorks provided us with the technology and professional services that
were critical in developing a prototype on a 2.5G wireless network. We also
believe that the multimodal functionality will be an attractive application
for sales force automation and other end user applications such as driving
directions and personal information management.”


IBM said its new product, which contains a pre-packaged dictionary of more
than 160,000 words and specialized vocabulary sets available for specific
industries, allows professionals such as doctors and lawyers to dictate
medical records or legal documents any time, anywhere for automated
transcription.


Because it is server-based, users can train the system to recognize their
particular speech patterns, and accents, and then dictate from any location,
using any communication device. And the Topic Factory, a built-in tool,
allows developers to add specialty vocabularies, IBM said.

News Around the Web