How industry-leading accuracy and a variety of use cases make Axon Auto-Transcribe an invaluable tool for law enforcement
Police departments are dealing with more raw camera footage than ever before. From body-worn cameras to vehicle dash cameras, interview room cameras, and more, the sheer amount of video and audio created throughout a typical case can take days or even weeks to manually review and transcribe.
Complicating matters is the fact that environmental conditions can make video and audio footage recordings “noisy” and difficult to understand. Particularly with bodycam footage, which is often recorded in highly dynamic environments, factors like wind, sirens, traffic noise, or multiple speakers can make transcribing that much more difficult.
Developed specifically for law enforcement and tested using law enforcement data, Axon Auto-Transcribe provides fast, accurate transcription, whatever your use case. Whether you’re reviewing evidence, conducting performance reviews, or creating court-level transcripts, Axon Auto-Transcribe is the best, most accurate automatic transcription softwareavailable for law enforcement personnel.
Why transcribe?
Transcripts are the backbone of the evidence review process, essential for finding facts quickly and, in some cases, are required by law. Given the volume of video and audio footage available to officers, detectives and attorneys, the ability to quickly review footage to find the specific moment an incident occurs or when a witness said a specific thing is crucial. Having a transcription alongside the recorded footage allows law enforcement personnel to quickly ascertain the facts, saving critical time.
Not to mention that in some jurisdictions, 100% accurate transcripts are legally required. In these jurisdictions, automatic transcription software can save busy transcribers hours out of their day by providing starting points, rather than forcing them to begin with raw recordings.
In other jurisdictions, transcription might not be accounted for at all, leaving officers and detectives to transcribe their own interviews when they have time. Or a transcription may suddenly be required when no transcribers are available. In these instances, Axon Auto-Transcribe can be like an extra member of the team, providing fast and accurate transcriptions automatically.
Axon Auto-Transcribe also allows law enforcement officers to transcribe multiple videos simultaneously, something that would typically take much longer when performed manually.
“It allows you to do things that just weren't feasible with humans alone before,” says Noah Spitzer-Williams, Sr. Principal Product Manager at Axon.
Accuracy and ethics matter
Transcribing audio from something like a podcast or a video call is easy. Transcribing audio recorded from a body-worn camera in the field is hard.
“Body camera audio is some of the hardest out there in the world to accurately transcribe,” says Spitzer-Williams.
In the field, you have street noise, wind, sirens, people talking at different distances from the officer and all manner of regional accents. Different devices also have different capabilities when it comes to recording audio. The microphone on a body-worn camera might be different from the microphone on a dash cam. Or an interview room might be made of concrete and outfitted with a 20-year-old recording system, making the audio sound terrible. Accurately and efficiently transcribing from all of these different kinds of audio is challenging, but critical.
That’s why Axon has thoroughly tested Axon Auto-Transcribe specifically on law enforcement data. This ensures that Axon’s solution is the best auto transcription software for law enforcement. Unlike other solutions that may have been tested and developed for other, less demanding situations, Auto-Transcribe was developed and tested where law enforcement lives: on the streets, in the interview rooms, and on the dashboards of cruisers.
Axon has also sourced data from different regions, utilizing audio from people of different backgrounds, ethnicities, sexes, and geographical regions to ensure that Axon Auto-Transcribe can accurately transcribe audio no matter who the speaker is or where they’re from.
“We source data from all over,” says Spitzer-Williams. “We source data from all over the country across ages, accents, genders, and races. Those are the four dimensions. That's what allows us to benchmark our performance and make sure that there are no dramatic differences between those dimensions. Meaning, for example, we don't want men transcribing significantly more accurately than women (it turns out female voices are slightly easier to transcribe).”
Use cases for automatic transcription software
Axon Auto-Transcribe was designed to provide accurate transcriptions within five minutes, which makes it an invaluable tool for a variety of use cases.
“We will transcribe the file regardless of how long it is. Even if it's 10 hours long, we'll have it within five minutes,” says Spitzer-Williams. “To many users, it probably literally feels instant because we transcribe most data as soon as it’s ingested. By the time they actually pull up the evidence, the transcript is already there.”
Fast Evidence Review
One of the primary use cases for Axon Auto-Transcribe is Fast Evidence Review. In this scenario, an officer or detective will have the video or audio recording, and alongside that will be the automatic transcript. This allows the officer or detective to quickly skim through the transcript to find the moment on the recording they’re looking for.
With Axon Auto-Transcribe, you can skim the written transcript – which in many cases will be faster than watching a video – or you can use keyword search to quickly find the time code of the spoken word or words you are looking for.
“Fast Evidence Review has been making it easier and – at least anecdotally – faster than reviewing evidence the old school way,” says Spitzer-Williams.
Priority Ranked Video Audit
As part of the regular performance review process, supervisors are called on to watch video recordings of officer interactions. To eliminate bias, these reviews are conducted randomly. The supervisor can’t choose the specific person or video to review, they’re simply presented with body-worn camera footage of a random incident or interaction.
The challenge with these reviews is there are a lot of videos to sort through, and any given interaction might be fairly typical, leaving the supervisor little to offer or causing them to spend time on lower-priority interactions. Priority Ranked Video Audit allows supervisors to sort these random videos by keywords so that they can see the interactions that might contain hate speech or other flagged words and review those immediately. This allows supervisors to quickly get to the videos that demand their immediate attention, saving time and effort.
“The outcome there is that we're finding more opportunities for highlighting great officer behavior and areas for improvement,” says Spitzer-Williams.
Redaction
Body cameras and other recording devices capture a lot of data. Some of that data isn’t suitable for public release or needs to be protected for legal reasons, such as the names of minors or a person’s home address. Typically, police records are redacted after the fact, through the laborious process of someone physically entering the record and redacting specific words.
For video or audio files, the process is even more cumbersome. Someone must review the recording and manually mute out the word or words that need to be redacted.
Axon Auto-Transcribe makes redaction easy by redacting the word in the transcript and muting the word in the recording – automatically.
“What we've done is we've supercharged the redaction experience by bringing in transcripts,” says Spitzer-Williams. “So now you can find a word and then just click it and say ‘mute that word.’ And we will automatically mute it – because we know literally to the millisecond where we should apply that mute.”
To learn more about Axon Auto-Transcribe, visit the Auto-Transcribe product page or contact us to speak with Axon directly.
:format(webp))