Sponsored by Nuance Communications
By James Careless for Police1 BrandFocus
Police reports are the bedrock of the justice system. The notes officers take at incident scenes and then compile into reports afterward, establish the basic facts of the criminal cases presented to district attorneys (DAs). The DAs’ decisions on which cases to prosecute is tied directly to the police reports they receive from arresting officers.
Unfortunately, the timeworn practice of officers writing notes by hand and then typing reports hours later is fraught with potentially serious downsides. Officers on the scene are often under pressure to write the quickest, most condensed notes they can to save time. Key details can get left out, and writing can be hard to decipher after the fact.
Back at the station, writing a report based on these condensed notes and the officer’s personal recollections leaves the door open for inaccuracies. (Add the fact that most officers are not trained typists, and there’s even more room for error!)
Researchers have found that memory fades fast: Within one hour, people will have forgotten an average of 50 percent of the information presented to them. Within 24 hours, this jumps to 70 percent, and within a week, forgetfulness reaches its peak of 90 percent. Working from recorded notes can help but transcribing these notes by hand eats up precious time.
With pitfalls such as these, conventionally prepared police reports can have unintentional errors in them. At best, an officer who already could barely spare the time needed to create the report has to go back in a second time to correct an error. Uncorrected, that error can be repeated down the line as a case proceeds to court. Simple mistakes such as misspelled suspect names, wrongly noted license numbers and other typos could impair the DA’s case if unveiled in court.
As retired New Jersey police division commander James Filippello commented on the popular Quora website, “An inaccurate police report is a special gift to most defense attorneys.” Worse yet, “Most officers will make the inadvertent error in a police report which will many times go unnoticed by the officer’s reviewing superior,” he wrote – until it gets discovered and exploited by a defense attorney during a trial.
In extreme cases, the consequences of inaccurate police reports include allowing guilty suspects to go free and the public to be put at risk. There’s also the potential for reputational risk to the report writing officer and by extension the department; all because the officer (who never joined the force to be a typist) made an honest, inadvertent error.
Speech recognition improves report accuracy
To improve police report accuracy while documenting an incident many police departments are moving to speech recognition solutions such as Nuance’s Dragon Law Enforcement.
Using Dragon’s next-generation speech engine powered by deep learning technology, the software allows police officers to create incident reports by dictating directly into their department’s records management system (RMS), either on their vehicle laptop or back at the station. The Dragon platform also accepts audio notes recorded on the officers’ smartphones or handheld digital recorders and will transcribe these spoken notes into text that can be incorporated directly into their reports.
Dictating incident reports by voice helps prevent errors and inaccuracies because Dragon Law Enforcement supports correction commands and incorporates a “play that back” feature allowing officers to hear their report read back in their own voice prior to submitting it. This empowers officers to format and file highly accurate reports in compliance with departmental standards.
Chief Kyle Heagney of the Attleboro, MA police department has watched this dynamic play out in the past year. “Since adopting Dragon Law Enforcement, my officers and detectives are capturing the immediacy of an incident or crime with greater clarity and specificity. And the ability to ‘proofread’ the document by having it read back allows officers to quickly spot the occasional discrepancy between what they spoke and what is transcribed. This has led to higher-quality reports to prosecutors and the DA.”
Speak the language of law enforcement
There are general-purpose speech recognition products on the market today, but they lack the police-centric focus of Dragon Law Enforcement.
For instance, Dragon Law Enforcement is built with a language model and dataset that includes words and phrases uniquely encountered by law enforcement. That means the software
accurately transcribes spoken conventions – like the common usage of the NATO phonetic alphabet (“Alpha” “Bravo” “Charlie”) – into text, without the misinterpreted errors a general-purpose speech recognition platform may incur.
Dragon Law Enforcement is so attuned to police needs; the software allows custom words to be added to its dataset by authorized police IT administrators and “pushed” to the entire force to instantly be available to all officers. This is useful for adding unique sounding street names within the community. Even the unfortunate reality of having to document profanity spoken by a suspect has been anticipated, with over 450 words of profanity built into the language model.
Mindful that police deal with a lot of vehicular crimes, Dragon Law Enforcement has most makes and models of cars and trucks instantly available for officers. When an officer speaks she has pulled over a “1994 Toyota Camry,” Dragon understands and formats the vehicle make and model correctly. And the software can accurately understand words spoken using a wide range of accents, even in noisy environments.
Returning more time for active police-work
Using Dragon Law Enforcement allows police officers to prepare reports that are more accurate than hand-written/typed reports, and in far less time.
More accurate reports mean far fewer errors, and thus fewer potential consequences for officers, the department, and the public.
Less time spent preparing reports means more time available for active police work, keeping the community safe, and affording more time for community policing initiatives.
This last point matters: Nuance’s research shows that police who use conventional report-writing methods spend 25% of their shifts doing paperwork.
When speech recognition software, like Dragon Law Enforcement, is incorporated into a department’s documentation workflow, it can be a de facto “force multiplier” for police departments large and small by reducing the amount of time an officer spends on reporting. This means officers can spend more time keeping a watchful eye on the community and fighting crime.
About the Author
James Careless is a freelance writer with extensive experience covering law enforcement topics.