Acquisition setup recommendation: insect monitoring

Question

I'm building an audio-based insect / pollinator monitoring device for crowdsourced biodiversity assessment.

I want to record sounds of bees, bumblebees and other pollinators arriving, departing and flying close to a flower. These events will be then detected and insect species automatically classified.

An ideal setup should:

• be independent on flower density; this means clearly capture sounds in a small volume close to the flower
• record an insect in the same distance from the microphone with the same intensity in different devices

I have made sample recordings using Zoom H1 pointed directly to a flower from ~ 5 cm distance. I have struggled to record anything usable with the cheapest clip Hama MicroClip CS-460.

The used Zoom H1 setup is workable, but the final setup should be suitable for a crowdsourced effort. Most probably a cheap microphone plugged into a smartphone.

• What microphone type / model / polar patterns do you recommend?
• What orientation and positioning? Should the microphone point upwards to reduce the influence of the neighbourhood?
• What inconsistency could be caused by different pieces of the same microphone model?
• What inconsistency could be caused by different smartphone model (different internal audio circuitry)?

Answer 1

have you considered using a contact microphone attached to the flower stem? Some sort of mems super-minature mic? Polar patterns imply large capsules. Directional properties are obtained using passive filters which are physically large in respect to the capsule itself. Best to use a basic omni mic and achieve separation through proximity to the sound source. Forget using a directional mic for this - it will end up being too big for the solution.

To get this to work, you will need to train an AI with multiple examples of particular insects in different situations with different mics, on different flowers.