Yowiehunters Forum

Interesting article about the "Acoustic observatory to record Australian animals and habitats"

Which had me thinking

They are going to be deploying about 300-400 sensors out in the Australian Landscape, some most probably in remote areas of bushland for months and up to a year, Will they happen to come across any kind of unknown sound(s) such as Yowie?

"Hundreds of audio recorders are about to be installed in regional Australia to create the country's first soundscape, which can give scientists important insights into different animals and their habitats."

"The acoustic soundscape will be available by early next year."

http://theconversation.com/world-first- ... alia-88306

https://www.abc.net.au/news/2018-12-18/ ... e/10631256

Hey Dion
I'll just bet that any recordings that dont fit into the normal slot will just be tagged as an anomaly and be wiped .
Research and science should be about the discovery of new things however with political and funding restrictions all the good stuff will be consigned to the waste bin !

Hey AL

I guess where I am coming from is, it would good for them to hear something that they couldn't put a finger on it source like a roar or howl other than being a possible wild Big Cat. I dont think they would wipe anything considered as a discovery of sorts.

I just feel for the poor bugger(s) that has to go through all the recordings of 300 -400 sensors and sort it all out into differing species.

Yes
To true Dion it is a job for the apprentice no doubt Lol
Like sorting fly poop out of the pepper

Dion wrote: Wed Dec 19, 2018 1:09 pm
I just feel for the poor bugger(s) that has to go through all the recordings of 300 -400 sensors and sort it all out into differing species.

Imagine reviewing all that data. Expensive and labor intensive operation.

AustralopithecineOz wrote: Wed Dec 19, 2018 4:53 pm Imagine reviewing all that data. Expensive and labor intensive operation.

Hey Dave , I believe they use cluster analysis . The computer analyzes the audio files .

The recent audio recorder I mentioned has an accompanying program which also has cluster analysis .
https://www.wildlifeacoustics.com/produ ... kpro5-land

Impressive Rusty.. in fact I'd call it a game changer.

Rusty2 wrote: Wed Dec 19, 2018 5:04 pm I believe they use cluster analysis . The computer analyzes the audio files .

The recent audio recorder I mentioned has an accompanying program which also has cluster analysis .

Hey Rusty

Forgive my ignorance but isn't cluster analysis just clustering certain wavelengths, wouldn't cluster analysis need someone to listen to the audio anyway to distinguish between species of animal? and also to make sure your listening to the right animal?

BTW wouldn't mind one of those audio recorders you linked to.

Hey Dave , yes it's a game changer and could be used for our benefit if only I could understand the mathmatics behind the analysis .

Hey Dion , from what I can tell , the people behind the algorithms and cluster analysis already have the known signatures of target species and calculate that into the program . The program flags audio files that match the various algorithms/species for review . Then yes , someone has to listen to each flagged file but it saves manually looking through an entire audio file . It is looking for certain wavelengths but also individual signatures in that wavelength based on Hidden Markov Models , it's complicated and It isn't 100% accurate .

Below is a link to white papers that talk about automatically identifying animal species from their vocalizations .
https://www.wildlifeacoustics.com/resou ... sentations

Will these audio recordings be available to the public online? I hope so because we can then compare our recordings to the animals they record.

yowiedan wrote: Sun Dec 23, 2018 7:01 pm Will these audio recordings be available to the public online?

"And once we begin recording, every minute will be made available to everybody online."

Sounds like it although as Al has mentioned , there may be unknown recordings that the public may not get to hear .

I downloaded the Kaleidoscope software ( https://www.wildlifeacoustics.com/produ ... kpro5-land)
for a trial period and have been seeing what it can and can't do .

In one particular mode it will collect all bird calls in a audio file and sort them into clusters . Some clusters may represent variations of calls from the same bird . At night time it will collect frog and bat calls .

For example , a software search through a 1 hour long piece of audio produced 10 clusters and 1300 seperate calls . It took an hour to look through them .
Cluster 00 produced calls from a gillbird . Cluster 01 , 02 , 03 and 04 produced clusters of variations of gillbird calls . There was a wallaby that hopped past the audio recorders in cluster 00 . The program did collect it but only because of the gillbird call at the same time .

Clusters 05-09 produced calls from different birds .

Here's the problems for anyone who wants to use this to save time .

The software may very well detect woodknocks , vocalisations and calls (sounds) but it's looking for birds and frogs and may overlook anything that doesn't fit into the parameters .

The software can be trained to look for odd audio but you have to have audio files which contain odd calls/sounds so that it can learn what is what .
We know there's different types of calls , woodknocks and unknown vocalisations so if the software doesn't know what to look for , it won't be able to see the important stuff .

It may be handy to scan audio for sounds that may interest us but ultimately it will miss something . It's a very cool tool and looks like it's a tool for surveying bird and animal life in a particular area but at $399 U.S. a year or $1500 for a permanent license I won't be buying it .

It takes me around 3-4 minutes to look through 1 hour of audio with Izotope Rx . Izotope Rx is much better value for money and at $299 U.S. is probably the best tool I've ever bought . Not only can you immediately see vocalisations and sounds of interest you can also enhance and repair sounds that might need attention .
https://www.izotope.com/en/products/rep ... 07_iZotope

Rusty2 wrote: Thu Dec 27, 2018 9:01 am I downloaded the Kaleidoscope software ( https://www.wildlifeacoustics.com/produ ... kpro5-land)
for a trial period and have been seeing what it can and can't do .

It may be handy to scan audio for sounds that may interest us but ultimately it will miss something . It's a very cool tool and looks like it's a tool for surveying bird and animal life in a particular area but at $399 U.S. a year or $1500 for a permanent license I won't be buying it .

It takes me around 3-4 minutes to look through 1 hour of audio with Izotope Rx . Izotope Rx is much better value for money and at $299 U.S. is probably the best tool I've ever bought . Not only can you immediately see vocalisations and sounds of interest you can also enhance and repair sounds that might need attention .
https://www.izotope.com/en/products/rep ... 07_iZotope

Sorry for a late reply been away from the forum for a while.

That Kaleidoscope software sounds awfully expensive and if the same thing can be done with izotope, well why waste the money.

Good tool for a survey Dion but that's about it .