A quick look around the just opened Flatbush, Brooklyn location of
A new voice-recognition technology appearing in wireless phones is said to combine the best of handset-based and network-based voice-dialing technologies.
Depending on implementation, developer VoiceSignal Technologies said its technology can also be used to voice dial by digit, launch phone functions by voice command and dictate messages that are automatically converted into text messages,
So far, the technology's voice-dialing and voice-activation functions are available in the Samsung's Sprint-network a500, but it will be available in June in two other Samsung phones: the i700 PocketPC PDA-phone for Verizon and the a600 for Sprint. By the end of the year, the technology will appear in nine to 10 additional handsets throughout the world, mainly in Samsung models but also in Nokia's 60 series Symbian-based PDA-phone, said VoiceSignal CEO Daniel Roth. Phones destined for the United States this year include a Samsung-made Verizon flip phone that will incorporate Microsoft's Smartphone 2002 OS, formerly called Stinger.
Next year, Roth expects the technology to appear in select Palm-based PDA-phones.
Before VoiceSignal's technology appeared, handset makers relied on older voice-tagging technology, which enabled users to dial 10 to 30 contacts by speaking the contact's name into the handset microphone. To program the names into memory, consumers record the names in their own voice and tag each of these "voice tags" to specific phone numbers in the phone's contact list.
VoiceSignal's technology, in contrast, eliminates the manual voice-tagging process by automatically creating a voice entry for each contact in a contact list. As a result, every name in a phone's contact list, not just 10 to 30, can be name dialed. In addition, any number — including numbers not in the contact list — can be digit-dialed by speaking the numbers, Roth said.
The phones are shipped with synchronization software to transfer a PC's Outlook contact list to the handsets.
Users can set the voice-dial feature to automatically activate when a flip phone is opened. They can also program a hard button to activate the feature, press and hold the talk button, or press a hands-free headset's talk button.
To dial by name, users say "name dial." To dial by digit, they say "digit dial." The phone then uses a synthesized voice to repeat the name or number and ask if the name or number is correct. The user then says yes or no.
Network-based voice-dialing services also enable digit dialing and name dialing of an entire contact list, which has to be uploaded to a network server, said Roth, but users have to hit three buttons to launch the service. In addition, subscribers who roam onto another digital network or to an analog network lose access to the service, he said.
VoiceSignal technology, Roth said, will help boost airtime usage by making it safer to dial a phone in a car. "Sixty percent of calls originate or terminate in a vehicle in the U.S.," he said.
The technology will also boost usage of other network features that require multiple keystrokes to access, he claimed. Consumers could say the word "browser" to launch a phone's microbrowser, then say the name of the Web site to access. To download games, consumers could say the words "play games." Carriers, he said, "are frustrated that all their technologies are hard to get to from menus."
The technology can also be used to call up a contact list and activate the voice memo feature.
In Korea, one phone uses the technology to provide talking caller ID, and it might appear in phones in the U.S. late this year or early next, Roth said.
Also in the fourth quarter, a U.S.-market phone will use the technology to provide speech-to-text conversion for sending text messages. A dictionary list will enable users to correct words that were interpreted incorrectly by the phone.
The technology uses so little processing power and memory that phone makes don't have to modify their designs to incorporate it, Roth claimed. About 100-150K of memory is needed for digit dialing, automatic name dialing, and command and synthesized speech functions, and 600-700K is needed for voice-to-text message conversion, he said. It also doesn't have a big impact on battery life, he claimed. "Network-based solutions draw more power because they're using the radio, DSP and CPU," he pointed out.