The year's best news, analysis and opinion sources for state and local tech leaders.
When city administrators drastically cut the Redlands Police Department’s budget last year, one of the first things to go was the $96,000 line item for dictation and transcription services. Suddenly, officers in the California community were stuck pecking away at their keyboards to finish critical reports.
“We had been using a service for almost seven years where officers could dictate reports into the phone and receive a transcription, and everyone was used to it,” says Lt. Travis Martinez, a Redlands police officer. “Once it was gone, officers had to spend more time typing, meaning less time on investigations and patrol.”
The service cut occurred at the same time the city downsized staff, making it even more detrimental to have officers tied up with paperwork. But Martinez found an affordable solution in Nuance Communications’ Dragon NaturallySpeaking voice recognition software.
Voice recognition software creates a profile that contains information about the unique characteristics of each person’s speech, along with a vocabulary of customized words. Officers then speak instructions to write reports, interview notes, search warrants, e-mail and other text-based documents. In addition to Nuance’s stand-alone offering, voice recognition software has been blended into many everyday tools, including Windows and Mac operating systems.
Dragon has worked so well for the Redlands police that Martinez purchased enough licenses for the department’s 76 officers and support staff.
Martinez says that voice recognition software has been a godsend in terms of time and money. If an officer takes someone into custody, he has to turn around a report within 24 hours. Dictating is much more efficient and saves on the overtime that might have been incurred to finish such paperwork, he says.
Dictation also leads to more detail-rich accounts because officers can verbally describe the scenes they encounter. “Not having to type results in officers being more thorough,” he says.
Patricia Kutza, a research analyst at BCC Research in Wellesley, Mass., considers public safety, including police and fire departments, a natural fit for voice recognition software.
“These departments have a special vocabulary that requires the use of very robust software like Dragon,” she says.
Voice recognition offerings have matured and are expanding to address the growing mobile-device market, according to Kutza. “Earlier uses of speech recognition technology were marred by high error rates and susceptibility to ambient noise. The technology has really improved in this area,” she says. In fact, Nuance claims accuracy rates as high as 99 percent out of the box.
In Florida, the Clearwater Police Department patrolmen use Dragon NaturallySpeaking to complete paperwork in the field, according to Officer James Perz. “Some of our officers are responsible for five to 10 pages of reports in a day. That’s a lot to write, especially when you’re sitting in a cruiser,” he says.
The officers can call up Dragon on their in-car notebooks and dictate their reports in a comfortable position. They can have Dragon read back the reports to proofread in case the trained dictionary did not get words correct. “Curse words, street names and people’s names can be problematic for the software,” Perz says. However, in those instances, officers come up with another word, use it consistently, and then find and replace that word with the actual one later.
Because the software works with the department’s computer-aided dispatch system, patrol officers can also use Dragon to quickly run license plate number checks as they’re driving.
Clearwater has 27 licenses for the voice recognition software, though Perz intends to upgrade to Version 11 soon and double the department’s license count. “Why wouldn’t I?” he asks. “It reduces an officer’s report-writing time by well over 50 percent. With its ability to get officers back on the street, it’s a great asset for any agency,” he says.
Public safety officials say the key to getting the most from a voice recognition software investment is to take the time to train users.
Lt. Travis Martinez of the Redlands Police Department in California advises his officers to speak slowly and clearly to get the best results. He concedes that not everyone is going to benefit from the software, such as those who are fast typists, so he focuses his instruction efforts on those willing to learn.
Officer James Perz of Florida’s Clearwater Police Department has a similar strategy and plans to develop a four-hour lesson to accompany the department’s eventual upgrade of Nuance Software’s Dragon NaturallySpeaking.
He points out that users must be patient because the program has to learn each officer’s voice style. He helps officers develop their vocabulary libraries with common terms upfront so they don’t have to spend time teaching the system later.