I see you :D

If, like Lars, you also want an animal borne audiologger group - drop a comment in the thread Lars and I have here: 

Voice activated recording devices on satellite collars | WILDLABS

Hi everyone,I'm cooperating with a project that will be placing satellite collars on Eurasian lynx and their prey species. I have a PhD student starting this year who is interested in vocalizations, and this is the perfect opportunity to attach voice activated audio recorders to the collars. Studd et al. 2021, did this successfully, but the recording devices they used are manufactured in Russia making them impossible to buy. I've seen publications that have used custom made devices on lions and hyenas (e.g., Wijers et al. 2020, Lehmann et al 2022), but I would prefer an off the shelf solution. I've looked at possible alternatives, but it is difficult to say whether they would be hardy enough to survive the cold and rain in Finland. In any case, I'm hoping that someone has some experience and I'd be very grateful if you can point me in the right direction.Thanks!John

wildlabsnet

Thank you for the reply Lars! 

Ideally it will be a general tool, with the configuration (biotic, abiotic, anthropogenic, etc.) decided by the user. But we are starting with a focus on biological sounds.

For rare/unknown sounds, we plan to utilize an on board acoustic clustering algorithm which will both identify rare acoustic events and avoid the recording of similar sounds. The requested acoustic datasets could be used to initialize the clusters. We are still experimenting with different approaches.

Cheers,

Jesse

Thank you for the comment Matthew!

The processing will indeed take power, but we are keeping the algorithms fairly simple and our system should be able to last over 1 year with a single C cell battery. You are right that some data will be lost with this approach - the tradeoff between longevity and detail! For the small size of the instrument, power and storage limitations wouldn't allow for 1 yr deployments with continuous data writing so it's more efficient for our system to do the processing. On board clustering will help ensure it is recording representative samples of similar acoustic events and we may 'prioritize' certain clusters for known acoustic events of interest.

Cheers,

Jesse

Just an anecdote here. And not audio. In the field of object detection for images small models means both more false positives and more misses.

Because smaller models means higher frame rates I experimented once with using the smaller model to detect intruders and if it got a hit it would then recheck with a bigger model.

However in the case of very small models they fired with false positives so often that there was no gain. I myself would expect the same with your situation with audio.

Also, I’ve run Lorawan based PIR sensors. They indeed can run for a couple of years on two AAA batteries. However they are almost doing no processing at all. They only process when they get an interrupt generated by a POR detector. There’s no way this would last for a year if the processor was running continuously. And there’s no disk writing going on.

How many micro amps does your processor use when running continuously with a 1.5v battery?

What processor are you intending to use ?

Thanks for sharing Kim.

We're using <1 mA while processing, equating to ~9 Ah running for a year. The battery is a Tadiran TL-5920 C 3.6V Lithium, providing 8.6 Ah, plus we will a small (optional) solar panel. We also plan to implement a threshold system, in which the system is asleep until noise level crosses a certain threshold and wakes up.

The low-power MCU we are using is https://ambiq.com/apollo4/ which has a built-in low power listening capability.

<1 mA certainly sounds like a breakthrough for this kind of device. I hope you are able to report back  with some real world performance information about your project @jcturn3 . Sounds very promising. Will the device run directly off the optional solar cell or will you include a capacitor since you cannot recharge the lithium thionyl chloride cell. I had trouble obtaining the Tadarian TL-5920 cells in Australia (they would send me old SL-2770s though) so I took a gamble on a couple of brands of Chinese cells (EVE and FANSO) which seemed to perform the same job without a hitch. Maybe in the USA you can get Israeli cells more easily than Chinese ones? 

Message me if you think some feeding sounds, snoring, grooming and heart sounds of koalas would be any use for your model training.

Really interesting project. Interesting chip set you found. With up to around 2mb sram that’s quite a high memory for a  ultra low power soc I think.

It might also be interesting while doing your research thinking about if there are any other requirements people could have for such a platform with a view towards more mass usage later. Thanks for sharing.

I love this list Dan, thanks for sharing!  

Thank you for sharing Dan! This is a great list. I'm really trying to find a labelled dataset for more than only birds though. Still a very helpful compilation of resources, and much appreciated.

The only non-bird, public, terrestrial acoustic dataset that I'm aware of is this one:

https://github.com/soundclim/anuraset