Let your fingers do the talking
In a digital communications system, voice is just a form of data: Speech is converted into packets for transport across the network and then reconverted to voice on the receiving end.
But in some communications systems, voice and data actually interoperate, as when a dispatcher transmits a text alert, and someone with a two-way radio hears the message spoken aloud, or when a user speaks commands into a wireless device to trigger a transaction such as retrieving voice mail, locating the nearest supervisor or joining a talk group.
Systems with some of these capabilities already are at work in public-safety agencies. Others, having found success in fields such as health care and field service, are starting to generate interest among first responders as well.
Codespear targets public safety with its SmartMsg system — often used with the company’s radio interoperability module — which provides a gateway for IP-based communications among incompatible radio systems, plus wireless and landline phones, computers, pagers and other devices.
SmartMsg lets a user at a computer type text or select a pre-formatted message and broadcast it to a variety of communications devices. For recipients who don’t have text displays, the system translates the message into spoken words.
“One message goes out, and it’s the same message no matter what type of device it hits,” said John Waters, a supervisor at the Livingston County (Mich.) Emergency Medical Services Department. “If you’ve got a device that talks, like a radio or phone, it just converts that to speech.”
Livingston EMS has implemented SmartMsg to help dispatchers contact first responders in the event of disaster across a nine-county district. With a couple of mouse clicks, a dispatcher chooses a prepared message, and “it goes out across cell phones, pagers, home phones” and other devices to a list of about 300 managers in police, fire and health departments, hospitals and other key organizations, Waters said.
Codespear uses text-to-speech capabilities built into Microsoft Windows, plus other off-the-shelf translation engines, to power its product, said Glen Seaman, the company’s vice president of business development and product strategy. SmartMsg can translate text to speech in 12 languages. It also can convert abbreviations and public-safety 10-codes into plain English for the benefit of elected officials and other non-specialists, Seaman said.
So far, Codespear hasn’t added a feature that translates speech into text. With the technology available today, converting a finite list of short phrases, such as “I need help,” for display on text devices is simple, Seaman said. But officials at Codespear are hunting for technology that understands anything a speaker might say, without any need to train the system to understand individual users.
“That way, instead of being limited to [simple] commands or … you can truly say, ‘I need to dictate the following all-call alert or the following directive,’ and it will be digitized and sent to all the members of a group,” Seaman said. “We just haven’t gotten there yet.”
Other developers, however, think translating a limited number of spoken commands into data is a fine idea. That’s the approach taken by Vocera, which offers a voice-over-IP communicator in the form of a wearable badge. Users speak through their badges to talk with others on a Wi-Fi network; they also speak short commands to execute transactions on the system.
“We try to resist, but it’s very tempting to talk about this as a Star Trek communications platform,” said Doug Glen, vice president of wireless solutions for BearCom, a radio system dealership that resells Vocera’s technology.
When BearCom demonstrated the Vocera system at the annual conference of the Association of Public-Safety Communications Officials (APCO) in Denver last August, it had been used mostly in hospitals, Glen said, adding that it could be useful to any public-safety agency that has access to Wi-Fi — be it a fixed network or a mobile network set up at the scene of an incident.
“But I think the more likely implementation would be within PSAPs and other fixed infrastructure,” where it offers an alternative to phone-based intercom systems, he said.
On the Vocera system, instead of using a keypad to enter an extension number, the user speaks a command such as “Call Bill” or “Find the nearest emergency medical technician.” The system can locate anyone who is logged into the system, based on which hot spot is picking up the communicator’s signal. Users also employ voice commands to leave and retrieve messages, start conversations among pre-defined talk groups, and set up, join or leave talk groups.
In addition, Vocera can translate speech into text for display on a small screen on the back of the badge. Users don’t seem to rely on that feature too much because the screen’s size doesn’t accommodate a complex message, Glen said. But the display is useful for storing details such as addresses and phone numbers as an officer responds to a dispatch, he said. “It complements the voice, but it doesn’t replace the power of voice.”
Vocera also plans to turn its communications device into an information systems interface. At the Healthcare Information and Management Systems Conference in San Diego in February, the company demonstrated the use of voice commands to query a database. This application employs technology from Nuance, which also provides Vocera’s speech-recognition engine.
Another proof-of-concept demonstration used Nuance technology in a transcription application, which would allow a health care professional to give dictation to the Vocera communications device. The system “would use speech-to-text technology to convert that into a document that could be attached, for example, to a patient record or stored as a memo,” said Brent Lange, vice president of marketing for Vocera.
Unlike Vocera, SandCherry doesn’t offer its own voice communications system. Instead, it piggybacks on existing two-way radio and phone systems in two kinds of applications that convert voice commands into data transactions.
SandCherry’s Voice4 Radio Messaging System (RMS) is an off-the-shelf application that lets people with two-way radios leave, retrieve and manage non-mission-critical messages. For example, if a dispatcher can’t reach someone in the field, he or she would record a message. Then “the individual comes back in radio range and can check this message, almost like voice mail,” said Dale Hartzell, vice president and general manager of SandCherry’s OEM and voice appliance business unit. “It saves the dispatcher or the field officer from having to track people down and generate all that extra radio traffic,” he said.
Tim Omland, network field supervisor for Qwest Communications in Rock Springs, Wyo., uses Voice4 RMS to improve communications among eight field technicians. He said the system increases productivity.
“They’re not sitting in their truck trying to get hold of somebody,” he said. “Once they try one time, if they don’t get hold of them, they just leave a message and then go back to work.”
Along with this packaged solution, SandCherry offers the SandCherry Voice Platform, which can be used to voice-enable data systems. Some public-safety organizations are testing applications based on this platform, although none were full-scale implementations as of press time, Hartzell said.
As SandCherry’s representatives pitch the system to public-safety organizations, customers are finding “there are two aspects of this technology that they like,” Hartzell said. “One is obviously the self-service because it doesn’t tie up the dispatcher, so people can get information when they need it.”
The other benefit, he said, is that a voice-controlled system boosts safety. Unlike a mobile data terminal or computer, “it allows them to keep their eyes up while they’re doing these transactions, so they can keep their eye on the suspect or the suspect vehicle and get the information they need.” Also, it allows an officer with a hand-held radio to retrieve information from a database while away from his or her vehicle, he said.
Accessing data through a radio is also a less expensive solution, Hartzell said. “Many departments can’t afford $5000 data terminals in vehicles.”
SandCherry uses speech-recognition technology from Nuance and Lumenvox and relies on Nuance to power its data-to-speech conversion. In both the messaging and data-retrieval systems, the computer running the SandCherry software is connected to a radio and operates like a node on the radio network. The software “listens” to that radio for “hot words” such as “license check” or “get messages,” which signal a data transaction. The applications rely on a limited set of voice commands.
In any voice recognition system, background noise is a concern, and it could pose a particular problem for public-safety personnel working on busy city streets. For the Vocera system, BearCom offers a variety of noise-canceling headsets. “Down the road, we’ll probably experiment with some throat mikes and things like that to get into a tactical environment, but we haven’t gotten there yet,” said Mike Butler, national project manager for BearCom.
Although background noise certainly is an issue, push-to-talk radio actually provides a good environment for speech recognition because it limits the time during which the microphone is picking up sounds, Hartzell said. Also, the high-quality microphones and speakers built into two-way radios help a great deal. If a radio signal is so weak that a human can’t understand a spoken message, “the system doesn’t have a chance either,” he said.
SYSTEMS THAT OFFER INTEROPERABLE VOICE AND DATA
|Voice-to-data capabilities: Not currently available|
|Data-to-voice capabilities: User types text alerts or selects pre-formatted alerts; system sends alerts to users via landline and wireless phone, pager, e-mail, mobile radio and other devices. If end-user device is voice-only, system translates message into voice and speaks it aloud.|
|SandCherry||Voice4 Radio Message System (pre-packaged application)|
|Voice-to-data capabilities: User speaks commands into cellular phone or two-way radio to retrieve and manage voice mail messages.|
|Data-to-voice capabilities: System automatically broadcasts machine-generated alerts to tell users they have messages waiting.|
|SandCherry||SandCherry Voice Platform (for custom-designed applications)|
|Voice-to-data capabilities: Used to develop applications in which users speak commands on radio or phone to conduct transactions with an information system.|
|Data-to-voice capabilities: When user retrieves data from an application, system translates the result into machine-generated speech and received by the user.|
|Vocera||Vocera Communications System|
|Voice-to-data capabilities: User gives voice commands on dedicated communications device to: call other users, locate others by name or job description, leave a message, call a talk group, form a talk group or call outside the system through the PBX. Spoken information can be translated into text, which is displayed on a small screen on the communications device.|
|Data-to-voice capabilities: Not currently available|