A lot has been going on lately. About two weeks ago, my contact in the Laurentians (aka my neighbour in Morin-Heights) has given me the signal I was hectically waiting for: The frogs are back. And they are millions.
I cancelled everything I could and drove up North as soon as I could. Time mattered: I didn't want to miss the frog season this year. It took me a few days to get back to it, find my old source code, verify and fix the circuits, etc. Luckily, my neighbor Jean-Luc was right: the warm days of early Summer had set the marshes on fire... at its pinnacle, right after sunset, the sound of the frogs was overwhelming, thickening the air, filling every channel.
Still, I had to bring several changes to the source code in order to get a better recording. The sound being too feeble, I had to amplify it software-wise by multiplying the signal. I also had to run a highpass filter to remove the very low frequencies that take a useless, large part of the audio spectrum. However, running these operations also increased the “aliasing” of the sound, making the “jumps” in the 8-bit signal more audible.
To put it simple: the sound was crappy.
I was able to enhance the sound quality by recording the full 10-bit data that I can get with the analog-digital converter (ADC) and applying the operations on this data before converting it to 8-bit. That allowed me to get a better sound quality and amplify it even more.
I ran some experiments using the ffft FFT library for AVR which revealed that (1) it's possible to plot the audio spectrum on the AVR in realtime and (2) there is a significant (visible) predominance of a part of the spectrum (in the high frequencies) when capturing the sound of frogs. However, at this step, I didn't have the time to do anything useful with those results in terms of behavior and reaction, so I relied on a more simple algorithm instead.
Finally, I made use of the sleep functionalities of the AVR combined with the watchdog to save energy during less active behavioral states. An example is available here although (1) I don't really understand where the original code comes from and (2) what the hell is the “nightingale code” this guy is referring to?
I was lucky to get the help of Jean-Luc who is really the “king of the forest” (at least around the area). He found the perfect spot for installing the objects and assisted me in installing the devices. I also need to thank Francine, Édithe and Étienne for helping me carrying the devices. We got everything installed on May 16th. There was some issues with some of the modules (one of the wires broke and one of them was working strangely) but the main effect was great. I still have some programming issues though, the rythm at which the modules are activated is too fast (they seem to be too sensitive to sound). However, the global effect was really up to my satisfaction.
I was happy to have into my possession the Cannon EOS 7D borrowed from Perte de Signal to take amazing pictures and video shots of the installation. I'll soon make a video montage showing the devices in action.A lot has been going on lately. About two weeks ago, my contact in the Laurentians (aka my neighbour in Morin-Heights) has given me the signal I was hectically waiting for: The frogs are back. And they are millions.