Goertzel filter for real-time classification

There must be hundreds of Audio Moths in the wild now, and perhaps there are alternative versions of the code in use, but not a word of it on this forum! Is there another forum we should be participating in?

My Moths are collecting dust waiting for me to learn a LOT more about programming for signal detection, without ANY examples to work from...

It's a year and a half since the code used in the New Forest Cicada project was to be released 'very soon now'. I appreciate that research priorities are obviously elsewhere, and we're all under the pump, but isn't it time that the raw code, warts'n'all was released so those of us who really need this? The Open Source community doesn't sit on code, there's a community of enthusiasts out here keen to understand and improve it.

(Apologies for the tone, but I'd REALLY like to get on with this. If anyone has suggestions for how we might create a crowd-sourced & funded alternative firmware, I'm in.)

has anyone seen the sourcecode or pointers for this it would be HUGE HELP to me and ecological projects, I'm happy to attribute sources and reference in papers

anyone ?

Hi everyone,

We are planning an open source release of several of our detection projects using AudioMoth imminently. This release will include documented code implemented for the New Forest cicada project, as well as some others to help any future users wishing to implement their own detection firmware.

The release for this will be very soon, we'll notify you all when they're on our public repositories.

Peter

We have also been looking at gun shot detection in a nature preserve in tropical rainforest. Our initial experiments have been looking at vectors of change in spectral centroid and Kurtosis and Amplitude. Initial trials used RP3's. I have not looked at the code or processor for the AudioMoth yet (still waiting for mine from the last round), but perhaps someone would know if there is sufficent DSP there to do such analysis? I did read in the paper above that you re doing continuous analysis using 3 buffers, so this peaked my interest in asking who else is looking at these challenges. We would hope to add event detection messaging using long range wireless developed for IoT applications so that gardians of the preserve could be alerted to poaching activity.

In the interests of assisting the development of the detection algorithms, I submit several URLs to the calls of the Rufous Scrub-bird Atrichornis rufescens, a species of significance in the rainforests of Eastern Australia.

https://www.hbw.com/ibc/species/57581/sounds & https://www.xeno-canto.org/species/Atrichornis-rufescens

It appears that the fundamental of these calls may be anywhere between 2-6KHz.

From the support forum it is clear that many of us are keen to implement recording triggers, and more than a few are ready to experiment with them in code. Perhaps the authors would refer us to literature describing the approaches taken, and share snippets of the code used in the New Forest Cicada project for us to experiment with.

Hi Andy,

Thanks for replying. Looking forward to your email once the code is restructured and available.

We are looking at the following:

(a) Detection of dynamite fishing - known to occur along some rivers here

(b) Detection of moving vehicles (diesel engined typically) - to make an event log of movements at remote locations. (eg. illegal sand mining is almost always with a lorry moving at night)

With (a) ideally we would need to be able to make an approximate location estimate. Perhaps some triangulation possibilities exist. But this is probably the next step.

Arvind

Hi Arvind,

Thanks for the message. We are doing a bit of code restructuring for the algorithm implementation for the latest hardware release, so the Goertzel filter code is not currently online. The aim of the restructuring is to improve the readability of the code and allow users to modify the detection code for custom applications with the current hardware. We will send out an email to all subscribers when the next software is released.

It would be great to understand your application?

Thanks,

Andy