For her PhD research, bat biologist Kadambari Deshpande made overnight recordings of bat echolocation calls in the Western Ghats. A “good night” would generate about 30 GB of data from 11 hours of recording with a bat detector. To process the data, Deshpande would go through several one-minute recordings, scanning every millisecond for bat calls, and make notes on the species and other information on their behaviour and ecology.
“It took me 11 months to process 20 nights of data,” Deshpande said. “BatEchoMon can probably give me that in a few hours.”
BatEchoMon, short for “Bat Echolocation Monitoring”, is an autonomous system capable of detecting and analysing bat calls in real-time. It is India’s first automated bat monitoring system, developed by Deshpande and Vedant Barje under the guidance of Jagdish Krishnaswamy, as part of the Long-Term Urban Ecological Observatory in the School of Environment and Sustainability at the Indian Institute for Human Settlements (IIHS), Bengaluru.
Deshpande is a postdoctoral fellow at the Observatory and the School; Barje, who leads the WildTech Project at the Wildlife Conservation Trust, is a consultant there.
BatEchoMon marks a new chapter for bat research in the country, according to Deshpande. The monitoring system allows chiropterologists — scientists who study bats — to “go beyond data processing and towards asking interesting questions about bat ecology”.
“Not only will it lead to a smoother workflow, it will help people transition to recording bats in different parts of the country, allowing us to gain more insights on the natural history and ecology of different bat species,” Rohit Chakravarty, a bat researcher and conservationist at the Nature Conservation Foundation, said.
“I don’t know of any device internationally with an inbuilt recording plus call classifying unit. If my knowledge serves me well, BatEchoMon is a milestone in bat research globally.”
The bat in the machine
BatEchoMon is more than a bat detector. Aside from a recording device, it includes components that can record, store, process, and analyse species-wise bat activity on the fly. Bat detectors are only the specialised recording devices that can convert the ultrasonic echolocation calls of insectivorous bats into audible sounds for humans. “In [BatEchoMon], Audiomoth, a popular low-cost ultrasonic detector, has been configured to work as an ultrasonic microphone,” Barje said.
BatEchoMon is programmed to activate automatically at sunset, when bats begin flying, and continuously listens and analyses audio. The device’s brain is a Raspberry Pi microprocessor, which processes the data captured by Audiomoth. “It first isolates bat calls from other ultrasounds, such as those of insects or anthropogenic and environmental noises. Then the peak frequency and structure of a bat call are analysed to match a known pre-trained model, which helps identify the bat species,” Deshpande explained.
“The system uses a [convolutional neural network] based algorithm to do this,” Barje added. The output from the device is a spectrogram — a visual representation of the frequencies of an audio signal as it varies with time — of all detected bat calls, along with audio recordings of the portions with just the calls. The system also generates statistical data on which species has been most active through the night, which species was active when, and so on.
“Earlier, all of this needed to be interpreted manually, after painstakingly combing through hours of data,” Deshpande said.
The Raspberry Pi and its associated processing components are enclosed in a box measuring 200 mm × 80 mm × 80 mm. Other auxiliary components in the device include a solar panel plus battery and a WiFi communication unit for power supply and data transmission, respectively. In the absence of the sun, the device can last for up to eight days, according to Barje.
BatEchoMon also has a modular design, and its battery, charging apparatus, and the level of automation and data relay can be customised to the space it is installed in. But the team was reluctant to reveal more about the setup process at this time.
‘Suddenly, it became possible’
Bat ecology and acoustics is a nascent field in India, with just a handful of bat researchers recording bat calls and analysing them for ecological studies. Global bat-call databases such as ChiroVox and Xeno-Canto have few recordings submitted by Indians.
Deshpande has been using bat detectors since 2008 and has observed their evolution worldwide. In Europe, she said, detectors equipped with the associated software and reference libraries have saved scientists a lot of time. Since then she has wanted to develop something similar, but customised for the insectivorous bats that are more common in India.
A chance meeting with a fellow researcher and engineer from her hometown in Nashik, Maharashtra, kickstarted the BatEchoMon project. “Meeting [Barje] just revolutionised the whole thought. Suddenly, it became possible,” Deshpande said.
The duo went through numerous iterations, trying different microprocessors, algorithms, and power solutions — a “big challenge”, according to Barje — before arriving at the current version of BatEchoMon. Their primary goal was to include all the desired functionality in a user-friendly, low-cost package.
The core system of BatEchoMon costs one-third of advanced detectors and similar systems, according to Barje. He did not wish to disclose exact numbers, however.
The main challenge
In the last few months, BatEchoMon has successfully completed pilot tests in an IIHS site in Nashik. The team plans to test it for longer durations and in diverse conditions as well as to beta test the device with select users outside the organisation.
“Everybody comes with their own experiences of acoustics and different species. So eventually, we would like our colleagues to test it for themselves and share their experiences so that we can improve on the existing system,” Deshpande said.
The primary obstacle for BatEchoMon is the limited availability of reference libraries for the calls of many bat species. “Currently, the system can identify six to seven common Indian bat species. Going forward, we would like to include as many bat species as possible,” Deshpande said.
They also said they hope the device in its present form will be able to identify the species commonly seen in urban, peri-urban, and human-modified forested areas. The main challenge is to create robust training datasets to make good detection models for different species — a challenge they hope to overcome by collaborating with other bat researchers.
Fortunately, collaborations among Indian researchers are improving because of initiatives such as the ‘State of India’s Bats’ workshop conducted by the Nature Conservation Foundation and Bat Conservation International, according to Chakravarty.
“One of the key knowledge gaps identified by participants at the workshop was the lack of a thorough reference call library,” he said. “We also need more funding to conduct surveys in different parts of the country. It will allow researchers to identify more species and gather more recordings for training datasets.”
Nikhil Sreekandan is an independent journalist.
Published – April 14, 2025 05:30 am IST