New Software Helps Scientists Identify Animal Sounds
Anyone who's trekked through the jungle can tell you about all the crazy animal noises emanating from every direction. From the riotous calls of howler monkeys to the bizarre boiling-teapot sounds that some insects make, the collage of sounds blending into each other like some avant-garde sound-art composition can be almost disorienting to hear.
Now, thanks to software, scientists have the ability to record massive amounts of audio samples from ecologically diverse areas and use technology to help in the species-identification process—a labor-intensive task, given the difficulty in deciphering the sounds in many hours of recordings.
The BBC reports that a team of scientists from Puerto Rico created software that “would make monitoring the life found in tropical ecosystems faster and easier for biologists.”
From the BBC:
Dr Mitch Aide explained: "If a researcher had 50,000 one-minute recordings, it would take 833 hours to listen to them just once, no pausing."
The new technology was much faster, he said, and could process the same amount of audio in about an hour.
The program, dubbed the Automated Remote Biodiversity Monitoring Network (ARBIMON), “provides a database and analytical tools that integrate file management, audio processing, sonogram evaluation, species-specific identification modeling, and verification in a web-based application.”
All the scientists have to do is set up an iPod in the field and have the iPod record audio samples through their application, and the application will send those files (all properly named) to the scientists’ base station, where they will be uploaded to the database management system.
In the researchers’ article published on PeerJ, the authors state that “A normalized open source database schema using the MySQL database system is the cornerstone of our web application.”
According to Popular Mechanics, the scientists are able to “train the computer to recognize particular sounds from their data,” thus enabling a much more efficient way to identify sounds and match those to different animal species.
If you want a thorough understanding of how the ARBIMON program works, you should definitely read the researchers’ PeerJ article in its entirety as it gives a full rundown of how the system processes audio data and discusses how the project is managed.
From PeerJ:
The centerpieces of the design are the sensors that acquire the data and the methods used to process the data, allowing our system to handle a variety of sensors, use different configurations of these sensors, and to create an efficient way to relate the data with the type of sensor and configuration.
If you want to be part of the project, you can go to the ARBIMON website to hear some of the uploaded sound files and upload your own sounds to be evaluated, as the researchers opened up their project to any user.
