Walking and texting can be dangerous, especially if you are trying to cross a road while listening to music.
Audio Aware – a new application due out in March – may provide an extra layer of safety for either the distracted or hearing impaired, by listening for sounds of danger and alerting the user.
One Llama, a technology start-up, is developing the machine-learning technology that mimics the way the ear works, and believes their programs will make it easier for smartphones and wearable devices to constantly listen for sounds of danger, reported MIT’s Technology Review.
One Llama says “Artificial Ear” is the cornerstone of their Audio Aware machine-learning system; extracting hundreds of audio (and musical) features from the surroundings. The company claims their machine learning techniques “are the closest thing technology has to flesh-and-blood judgment you can trust.”
Kurt Bauer, One Llama president and CEO, told TheBlaze depending on the quality of the sensor device, the app may be able to detect sounds the human ear could never distinguish — kind of the same way dogs can hear things humans cannot.
“Spectrally, we can certainly do that if the sensor device is good enough,” Bauer said. “For example if electric cars, you know quiet cars, these days they’re hard to hear and it causes certain problems. So imagine if there are ultrasonic systems (on the cars) that we can detect for people cruising around, that’s pretty cool, and a helpful little safety item.”
The app, planned for release in March, is programmed to detect sounds like screeching tires and wailing sirens to alert the user by interrupting music or vibrating. The application will “listen” in the background on an Android smartphone, and will be pre-programmed to recognize some common perilous sounds, but users will also be able to add their own sounds and share them with other people.
But Bauer said the interesting thing about machine learning, is that it’s hard to make it better than the human brain, when it comes to judgement.
“We’ve done a lot of work in speech patterns and we have data that shows that a test – (asking) men and women to determine whether a speaker has a low, medium or high voice,” Bauer said with a laugh. “The answers were all over the road – it has to do with opinion and it’s highly subjective. The machine can’t be better than that, but in those cases we want to be just as good as the human ear.”
MIT Technology Review reports:
Audio Aware will pique interest among makers of wearable gadgets, who could bake the technology into smart glasses, smart watches, and fitness trackers. In those devices, Audio Aware could do more than just be alert to dangers: it could monitor health conditions, workouts, or even locations by paying attention to the sounds you make and the noises around you. Bird watchers might want to use it to home in on the differences between, say, a male chipping sparrow and a dark-eyed junco.
Bauer explained the technology can already pick up sounds that the human ear has trouble distinguishing in noisy surroundings.
“In a really noisy environment sometimes it’s hard to pick out a particular sound … a really tricky one is breaking glass,” he said. “We are constantly tuning our system, so you don’t have too many false positives, but in the case of breaking glass we found that we were able to detect the sound of breaking glass and the machine picked it up when we couldn’t hear it ourselves.”
A user might think walking on a city street — say in New York City — might mean the user would get dozens of warning “hits” since sirens and dangerous sounds almost blend into the background. But Bauer says they have the ability to tune the sensitivity of the sensors to determine distance, etc.
Bauer said the app will be available on Android in March, but that’s just the first. “It will be available on all the likely suspects for smart phones over time. But this will continue to develop over time, and as wearables become more advanced and more popular, the warnings will be matched with the device.”
In other words, your Google Glass could present a warning note in the wearer’s field of vision, or monitors like a FitBit could give the user’s wrist a gentle squeeze to alert the user of danger.
(H/T: MIT Technology Review)
Follow Elizabeth Kreft (@elizabethakreft) on Twitter.