A Constantly Evolving Conversation

Using technology to monitor and protect the environment

Cellphones have become such an integral part of day-to-day life that researchers from the University of Illinois at Chicago have studied what they call ‘phone walking’ — visibly holding a phone without using it. “If one wants to be part of the constantly evolving conversation, mobile devices should be always ready to be used instantaneously,” the researchers say.

But human conversation isn’t the only sound that cellphones can record. Since 2014, the team at Rainforest Connection has been using cellphones outfitted with solar cells and extra microphones to detect noises associated with forest and biodiversity loss. Currently, their technologies are used in places such as Sumatra, Cameroon, and Peru to detect and identify chainsaws, motorcycles, and trucks — which indicate illegal logging, poaching, and smuggling, respectively. In Sumatra, the technology has had an immediate and positive impact. “For the first time ever, staff members on the Reserve found themselves able to confront illegal loggers live and in person — and shut down their operations,” the organization reports.

In an interview with DigitalTrends, Rainforest Connection’s founder, Topher White, explains how machine learning has helped in the efforts, allowing the systems to identify a far broader range of sounds, not just in real time, but in years’ worth of archived recordings. In Ecuador, the systems now help local rangers and biologists monitor several endangered species of parrots by providing real-time alerts and data analysis. White notes, “We should be able to detect animals that don’t make sounds. Jaguars might not always be vocalizing, but the animals around them are, birds and things.”

Machine learning is playing an increasing role in our efforts to monitor and protect the environment. Over at Google, the AI team has been working with NOAA’s Pacific Islands Fisheries Science Center to identify whale calls in thousands of hours worth of underwater audio recordings. “It would take over 19 years for someone to listen to all of it, working 24 hours a day!” project manager Julie Cattiau writes in a blogpost. To help, the team developed a deep neural network that learned to pick out the humpback whale calls, a problem that is particularly challenging due to underwater noise and the complexity of the songs. Once identified, however, the whale songs can be used to better understand the whale populations and to protect the animals by alerting ships that whales are present so that the vessels can go around them.

Today, machine learning has been applied to help identify and understand a wide range of creatures from birds to mosquitos, frogs to bats. It’s inspiring to see artificial intelligence used to monitor and protect the natural world, helping us sustainably manage forests and halt biodiversity loss. The next time you’re out phone walking, consider listening to the rainforest, too, courtesy of the Rainforest Connection’s app, which brings the sound of the forest to us, so we can listen to its constantly evolving conversation.