Speech recognition tools are changing the way people work and live their lives, bringing assistants like Alexa into their daily routines and transcribing meetings more accurately than a human could. When it comes to policing, speech recognition tools can help uniformed officers and detectives with critical note-taking and even possibly help prevent life-threatening situations.
ZDNet spoke with Mark Geremia of Nuance Communications and Chief Joseph Solomon of the Methuen, Mass. Police Department to talk about the way police work is evolving thanks to voice recognition, cloud computing and other new technologies. The Methuen PD is one of the police departments now using Dragon Law Enforcement, Nuance's voice recognition product designed specifically for first responders.
Here are some highlights of the conversation:
Possibly preventing police ambushes: Solomon said the biggest advantage of voice recognition tools is safety. An officer can sit in his squad car and dictate notes while still watching the streets around him. Solomon noted that this is especially important to officers, given the rise in police ambushes. Ambushes against police officers did indeed spike in 2016 and remained a concern in 2017. Additionally, the more an officer has his head up in his squad car, the more he's engaged with the community, Solomon explained.
"They literally can still monitor everything, which I think helps prevent ambushes on police officers and keeps that presence up in the streets," he said. "We didn't want officers writing reports in the cars because they weren't paying attention to their surroundings."
Voice recognition also allows officers to work more quickly and to take more detailed notes, Solomon said.
Tailoring voice recognition tools for law enforcement: Nuance builds voice recognition tools for different verticals, in part because different professions can have very different vocabularies, Geremia explained.
For instance, "profanity is a huge one that is enabled in this product set but not in others," he said. Additionally, Nuance works with each police department to learn how they pronounce commonly-used street names and surnames. They also customize the product so transcriptions come back with the appropriate abbreviations.
For instance, Geremia explained, Nuance will ask, "'How do you write 'sergeant?' Instead of typing up 'sergeant,' it types up 'Sgt.'"
Nuance also gives police departments the Nuance PowerMic, which cancels out ambient noise. "When you're in an automobile with the sirens on, the windows down and the AC blasting, [voice recognition] is still going to work, but you need the microphone," Geremia said.
Implementing new tech and upgrading IT: "Cops hate change more than anything," Solomon said. So when his department considers a new technology, they run a pilot. "We ask for volunteers, and I always say, 'Anyone who doesn't want this system, please volunteer.'"
Voice recognition was a natural fit because "cops are talkers," he said. A side benefit of adopting new tools like this, Solomon added, is that it forces a department to upgrade its IT.
"Historically police departments have legacy systems that are garbage," he said.
Cloud computing and data storage policies: The Methuen Police Department stores all of its body camera footage in the cloud with Axon, whose security and encryption have been vetted by the National Crime Information Center. Policies for storing such footage are largely dictated by state law, Solomon explained -- certain records have to be kept for a period of time. Still, the department at times sets its own parameters beyond the basic requirements. The department has auto deletion policies as well.
"Our goal is in the next 18 months to two years to move all our storage to the cloud, but that takes money, and I'm not sure the municipality has funding for that," Solomon said.
The future of voice recognition: Without committing Nuance to any roadmaps, Geremia said that the company is working on solutions that could help police departments leverage recordings like body camera footage.
"There are still so many recordings out there, we are working on technology that can parse that apart, that can listen to a video and pull out keywords," he said.
Solomon said he expects AI-based systems will soon be able to pull out valuable information from the vast amounts of data collected by law enforcement. That capability will make crowd sourcing information an especially valuable investigative technique, he said.
Public accountability: Laws to protect citizens will need to be changed as the technology used by law enforcement changes, Solomon said. But the key to staying accountable to the public is transparency, he added. "It doesn't work if you try to sneak around people... If you're honorable and only do what you say you're going to do [with the data collected], I believe there's no problem."