NICE Systems Enables Law Enforcement to Analyze Recorded Speech

With the recent spate of alleged police brutality in Baltimore, Ferguson, Mo., and other cities, law enforcement officials are looking for ways to share what happens from their perspective.

Similarly, emergency calls, recordings from interview rooms, and other critical communications contain information essential to expediting investigations and solving crimes, and allow for better training and responsiveness, according to John Rennie, general manager of NICE Systems' public safety business unit. Rennie sees the company's newly unveiled NICE Inform Audio Analytics as the solution for quickly finding relevant sound bites from these communications.

"This isn't just recording information; it's used to understand it quickly," Rennie says. The technology enables precise, fast searching and categorization of audio content based on keywords and phrases, with the analysis occurring immediately after recording or uploading (i.e., from body-worn cameras or other remote recording devices).

The technology is based on similar technology used in the contact center industry. But NICE designed Inform Audio Analytics to work in the noisy environments of 911 centers and in body-worn cameras, a "dirtier" environment than found in call centers, Rennie says.

"Our initial findings were that [the audio recording] capability that public safety people had did not meet their needs," Rennie says. There was no efficient way to quickly analyze, index, and search communications so that related calls could be grouped together. For example, a single shooting incident can involve different calls to 911 several minutes apart. But the callers typically will often speak to different dispatchers and could use different terminology. The Inform Audio Analytics technology can help determine the calls that are related, which can help the police in ongoing investigations, Rennie says.

The Inform Audio Analytics technology indexes identified words, assigning audio recordings to one or more user-specified categories, depending on the words or phrases.

N-Best indexing simultaneously matches words or phrases from a single audio recording with a confidence level for each word (e.g., "gun," 80 percent; "gone," 70 percent) to improve the probability of finding all occurrences of the target word or phrase, according to Rennie.

The technology can locate words and phrases faster than other audio search solutions that need to build phonemes into words and phrases for each search, according to Rennie, who says NICE Inform Audio Analytics can search through a day's worth of calls from a busy 911 center in just seconds. Such fast search capabilities are becoming more critical as 911 centers and police increasingly rely on audiovisual data, Rennie says.

Rennie says one large 911 center has used the technology for a few years, but, citing company policy, he would not provide any projections for future installations. Use in body-worn cameras has yet to progress beyond the proof-of-concept stage.

Rennie expects the recordings from body-worn cameras, 911 calls, and interview rooms to be used to help with:

• Investigations: to retrieve information faster or find calls that are part of a build-up to an incident, both of which can result in important insight to reveal potential witnesses and to speed research and resolution of an incident.

• Compliance: to ensure all protocols are being followed according to specific compliance requirements and identify non-compliant employees for remediation. For example, a police officer could have used inappropriate language during an arrest, Rennie says. With the technology, superiors could determine if the language use was a one-time incident or a pattern of behavior that needed to be remediated.

• Operational intelligence: to automate trend reports that highlight the types and quantities of calls taken during specific time periods.

• Quality assurance: to improve efficiency and effectiveness of 911 personnel by isolating specific types of calls for targeted quality assurance review; to automate call analysis to ensure that procedures are being followed and employees are consistently performing to the highest standards; and to identify gaps in training or policies that need to be addressed.

"We learn more from our mistakes than we learn from things going right," Rennie says. "People working in 911 centers have to be polite and accommodating. There is a large customer service aspect to emergency services jobs. This will help them provide a level of effectiveness and better efficiency for a better emergency services center."