.png){kind=logo}
Solving a crime is akin to putting together an elaborate jigsaw puzzle. Only once one can see how the pieces fit together do seemingly random acts start to make sense. Often when finding the pieces to the who-done-it puzzle, investigators have to think outside the box, especially when recreating key conversations or attempting to make sense of what they're seeing on security cameras or what's being said in witness accounts. In expanding their ability to reconstruct these vital moments, law enforcement agencies have turned to speech recognition technologies, especially text to speech and visual speech recognition.
More than not, investigators have to rely on witness accounts of conversations or bits of dialogue to piece together motives or potential suspects in a crime. Hearing these conversations second hand and without much context can make this difficult. Text to speech technology is making it possible for investigators to input dialogue and recreate the scene of a crime. Using human voices with accurate accents and tonal inflections can help those on the case to get a real sense of how a certain exchange may have gone down. Interestingly enough, text to speech can also be used to relay such conversations back in a courtroom setting, allowing the jury to hear out loud what transpired before or after a crime occurred. This can be paramount in making a case for or against a certain suspect.
It goes without saying that security cameras are a key piece of evidence in nearly every criminal case. Law enforcement agencies look at this footage straight-away, attempting to ascertain a suspect or piece together a timeline. Unfortunately, in most public settings, these cameras are not allowed to record sound. Leaving off valuable information that may have been exchanged in the form of a conversation or even criminal demands. Speech recognition is helping to bridge the gap between what is seen on a security camera and what can't currently be heard.
Lip reading technology developed by the University of East Anglia can be applied to any place where audio isn't good enough (or simply not recorded), to determine what those in the shot are saying. The technology works through the implementation of machine lip-reading classification technology. This advancement is able to differentiate between sounds thanks to visual cues, allowing for fully accurate translations. The ability of a machine to virtually read lips is invaluable to those in law enforcement. What's being said or murmured in security footage can supply a valuable context that otherwise would never be apparent or known. While the technology is still quite new, continued research and advancements in both text to speech and speech recognition will allow this field to flourish.
These advancements in the text to speech and lip-reading capabilities are amazing, but are they accessible to smaller precincts or agencies? Yes and no. While not every law enforcement department will have the funding available to utilize lip-reading software, most can easily access text to speech technologies online for free or for a small monthly fee. Big companies such as Google, Microsoft Azure, and Notevibes all offer free text to speech software with over 80 human voices and inflections to choose from. This is accessible technology that could just help crack cases big and small. When it comes to fighting crime, speech recognition is truly the next big thing.
