One of the best things about headphones is that they can shut out the noise around us and seclude us in our own little world, filling it with our favorite music or podcast instead of loud-mouthed bus riders or honking horns. The only trouble with shutting out the world is that our sense of hearing is critical for keeping us alive. One of the reasons we avoid letting buses or taxis plow into us while walking downtown is that we can hear them coming and stay out of the way. Now, One Llama, a startup that describes itself as “a machine learning company,” is developing a technology that mimics the human ear so that smartphones and wearable devices can perpetually monitor nearby sounds and warn us of approaching dangers, even when we’re drowning out the din with our favorite tracks.
According to MIT Technology Review report, the public will get a first glimpse at the potentially revolutionizing tech via One Llama’s soon-to-be-released app. The app won’t include the new tech in its entirety, but the New York-based startup hopes it’s app will garner attention from makers of smartphones, smart glasses, smart watches, etc. — the kinds of companies that could slide some cash One Llama’s way in exchange for permission to attach the software to future devices. The startup hopes the hazard-avoidance tech will eventually be a standard feature across a majority of smart devices.
Initially, One Llama will be available on Android only. It works by running quietly in the background, constantly scanning for danger via sounds like ambulance sirens, screams, tire squeals and more. The app will have a number of common hazardous sounds in its memory banks from day one. Here’s where the machine learning comes into play…
image credit: Sascha Kohlmann via Flickr
One of the app’s co-founders (and a researcher at the Illinois Informatics Institute), David Tcheng, explained that One Llama will listen through smart devices’ microphones and compare what it hears with its stored templates of alert sounds that it’s already able to recognize in the event of a dangerous emergency. When a close-enough match — a blaring car horn, for instance — is discerned, whatever you’re listening to will cut out, replaced by an amplified version of the danger sound. Tcheng added that the amplified sounds might even be cartoon-y takes on the sounds, providing easier in-the-moment recognition to the listener.
The technology is far from perfect yet. In light of this, users will be able to add their own “sounds of danger” to their One Llama banks, and even trade with other users to bolster their hazard databases.
You can expect to see the One Llama app available for Android soon, but there’s no set-in-stone release date yet. Until then, please just look both ways.
Editor’s note: This article has been altered to correct the name of the app, which was originally referred to as ‘Audio Aware.’ Audio Aware is actually the name of the patent.