Voice Match is for the Birds: New Google Competition Seeks Avian Audio AI
Google’s bioacoustics team and the Cornell Lab of Ornithology are looking for artificial intelligence that can detect and identify the sounds of birds. They are co-hosting a competition for models that could train an AI system to become an audio ornithologist, picking out and naming birds based on recorded soundscapes.
Cornell, Google, and their educational and conservationist partners are asking data scientists to take audio recordings from all over the globe and come up with a way to use machine learning to isolate the sound of birdsong and work out what species is making the sound. The audio used for the contest is curated from the crowdsourced sounds gathered by the Xeno-Canto project. Entries are due in September and the winner will earn $12,000, with $8,000 and $5,000 prizes for the second and third place winners, respectively. The contest takes place on Kaggle, a digital hub for data science competition with more than four million members worldwide.
“There are already many projects underway to extensively monitor birds by recording natural soundscapes over long periods,” Chrome media audio software engineer Tom Denton explained in Google’s announcement. “However the analysis of these datasets is often done manually, is painstakingly slow, and results are incomplete. Data science may be able to assist, so researchers have turned to large crowdsourced databases of vocal recordings of birds to train AI models. To fully take advantage of these extensive and information-rich sound archives, researchers need good machine listeners to reliably extract as much information as possible to aid data-driven conservation.”
The contest’s goals share some similarities with Google Assistant’s Voice Match feature. Voice Match is Google Assistant’s term for identifying individual speakers by recognizing their voices. The point is that there’s no need to switch to a different Google account to access personal calendars, messages, and other information if the voice assistant simply recognizes you. Google has been raising the feature’s prominence lately, including a more stringent setup process, making it more widely available on devices with Google Assistant, and even starting a pilot program for people to use it as an ID when making a purchase through the voice assistant. The new contest has a similar shape, except it looks to differ among species and finding those sounds among a much more complex audio landscape.
Birds are an incredibly widespread and diverse group. More than 10,000 species make their home pretty much everywhere that can support their lives. Because of where they stand in the food chain, environmental scientists use them to measure the health of a habitat and spot pollution. Birds are often the first victims of unsafe chemicals in the air or water. Coal miners sometimes used to bring a canary to work. If it passed out or dies that meant a dangerous gas was in the air, and the workers should leave. All birds are metaphorically a canary in a coal mine when it comes to pollution.
Collecting the audio of birds amongst a larger soundscape is difficult even for trained experts. A well-trained AI could use machine learning to mark and annotate the sound of every bird in a recording much more efficiently and accurately than a human over long periods. Do you hear the birds chirping outside your window? Over 10,000 bird species occur in the world, and they can be found in nearly every environment, from untouched rainforests to suburbs and even cities. Birds play an essential role in nature. They are high up in the food chain and integrate changes occurring at lower levels. As such, birds are excellent indicators of deteriorating habitat quality and environmental pollution. However, it is often easier to hear birds than see them. With proper sound detection and classification, researchers could automatically intuit factors about an area’s quality of life based on a changing bird population.
“With proper sound detection and classification, researchers could automatically intuit factors about an area’s quality of life based on a changing bird population,” Denton wrote. “If successful, winners of this competition will help researchers better understand changes in habitat quality, levels of pollution, and the effectiveness of restoration efforts. The eventual conservation outcomes could greatly improve the quality of life for many living organisms—birds and human beings included.”