24/7 Raspberry Pi/birdNET monitoring (1 Viewer)

[GMS]

Any tutorials on how to use birdNET analyser? I've installed it and it works fine when using short clips. When uploading longer .wav files the file appears not to be there on completion.

 

[Crex67]

Just bumping this as only a couple of days left on the kickstarter

 

[Buckster]

some interesting Spring Migration Stats for our Garden in Essex

Chiffchaff (maybe overwintering one) leading the way, followed by Blackcap (very active now), then House Martins (note March instance not correct, treat 6th April as first actual plot (I saw with my own eyes visiting nest front of house too), last week Swallows, and as of yesterday Lesser Whitethroat and today Garden Warbler (both quite active).

So far (over 4 months) - 93,000 bird recordings !

I've also included a plot of Med Gulls over last 2 months, I find with a lot of these recordings the changes fascinating, as in with the Med Gulls, big peak mid-April but hardly heard them since

made some tweaks to the BirdNET Pi - I've downclocked and undervolted the Pi 4b - so get around 26 hours with a 20,000 maH battery.

I also bought a cheap waterproof box for it - Masterplug Weatherproof Electric Box for Outdoors

 

[LittleBitOfBreadNoCheese]

Nice, have you detected any visitors you would otherwise have missed?

 

[Buckster]

Might well have missed:

Lesser Whitethroat - didn't hear myself but recording checks out - never heard in garden before either.
Garden Warbler - didn't hear myself but recording checks out - never heard in garden before either.
Overflying night birds - Moorhen, Coot, Water Rail - had no clue - have never heard myself.
Siskin - consistently detected not seen or heard personally from garden.
Snipe - detected but not seen or heard personally from garden.
Goldcrest and Firecrest - picked up daily - was not totally convinced as only 5-6 large conifers back of garden. I spent days and days looking and caught a glimpse of one of the other (was too quick to tell which) - so more convinced they are there I've just not seen.
Overflying waders, Dunlin, Avocet (good few times detected) etc

In the winter - Redwing - hadn't seen any - or heard any (but not been outside in garden for periods of time) - yet for a few months Redwing was in top 10 !

Missed first instances of

House Martin - it detected before we saw.
Swallow - it is detecting but despite looking not seen yet

 

[LittleBitOfBreadNoCheese]

Might well have missed:

Lesser Whitethroat - didn't hear myself but recording checks out - never heard in garden before either.
Garden Warbler - didn't hear myself but recording checks out - never heard in garden before either.
Overflying night birds - Moorhen, Coot, Water Rail - had no clue - have never heard myself.
Siskin - consistently detected not seen or heard personally from garden.
Snipe - detected but not seen or heard personally from garden.
Goldcrest and Firecrest - picked up daily - was not totally convinced as only 5-6 large conifers back of garden. I spent days and days looking and caught a glimpse of one of the other (was too quick to tell which) - so more convinced they are there I've just not seen.
Overflying waders, Dunlin, Avocet (good few times detected) etc

In the winter - Redwing - hadn't seen any - or heard any (but not been outside in garden for periods of time) - yet for a few months Redwing was in top 10 !

Missed first instances of

House Martin - it detected before we saw.
Swallow - it is detecting but despite looking not seen yet

Same story here on Goldcrest (we don't get Firecrest up here) and I've seen them plus the sonogram is quite clear.
Garden Warblers personally I need more reassuring, I heard an unfamiliar song in the local park the other day but I think it was what they call the 'long song' of a Blackcap. In the past birdNET used to struggle on splitting the two but hopefully it has improved.

 

[Buckster]

Same story here on Goldcrest (we don't get Firecrest up here) and I've seen them plus the sonogram is quite clear.
Garden Warblers personally I need more reassuring, I heard an unfamiliar song in the local park the other day but I think it was what they call the 'long song' of a Blackcap. In the past birdNET used to struggle on splitting the two but hopefully it has improved.

I believe BirdNET was correct with Garden Warbler and that my manual listening to confirm (you never know I could be in error too) is also correct

I did also check against Merlin

 

[Buckster]

I think this is better representation (first part of clip - with the downturn in tone at end of call)

can anyone confirm whether they think this one is Cettis please ? detected twice and sounds like it to me but would be unusual for my garden. First recording is followed by a Blue Tit. One recording in January, one in March

 

[GreenRooky]

I had common scoter over last night that I never would of picked up without the PUC

 

[Buckster]

I had common scoter over last night that I never would of picked up without the PUC

interestingly my BirdNET Pi picked one up yesterday too (I think) - maybe they are on the move

 

[Buckster]

90% sure (checked recording) - BirdNET PI was correct in picking up a Marsh Harrier call today (I've never heard one - seen loads though)

so over 4 months its managed recordings of the following raptors:

Barn Owl
Tawny Owl
Marsh Harrier
Red Kite
Buzzard
Kestrel
Sparrowhawk

pretty impressive

 

[GreenRooky]

BTO BBS 2023 has a few pages on how they are using BirdNet in the field and developing a UK specific classifier model

 

[Buckster]

very interesting thankyou for heads-up on that report

ref BirdNET PI - is it possible to change the recordings to 24 bit please ? I tried but failed

I tried changing all references in birdnet_recording script from s_16LE to s_24LE to no avail - just get very over saturated outputs

 

[LittleBitOfBreadNoCheese]

BTO BBS 2023 has a few pages on how they are using BirdNet in the field and developing a UK specific classifier model

The BTO science officer gave a talk to the SOC conference in November about his audio AI work but when I spoke to him he was a bit dismissive of birdweather on accuracy grounds. That was before the January probability model release though which I think has improved things at the birdNET level, plus Tim (birdweather guy) is pretty responsive if you have a specific (species level) issue.

 

[LittleBitOfBreadNoCheese]

very interesting thankyou for heads-up on that report

ref BirdNET PI - is it possible to change the recordings to 24 bit please ? I tried but failed

I tried changing all references in birdnet_recording script from s_16LE to s_24LE to no avail - just get very over saturated outputs

Don't have a Pi myself I'm afraid. Assuming you have hardware level support it might be best to raise it on the github Issues page.

 

[Buckster]

now up to 185k bird recordings with the BirdNET Pi

this was an interesting set of recordings from a few weeks back - be interested in people's thoughts

Woke up in the night and decided to look at the BirdNET Pi to see if any owls had been detected and saw over 50 instances of Magpie around 00:30 (which seemed odd - very late) interspersed with Tawny Owl - so thought I'd have a detailed listen to see what was going on.
Can't work out if group (or individuals) disturbed the other, territorial, both the Magpie/s and Tawny Owls were disturbed by a common foe (Fox, Cat etc) or one of the sets of the birds attacked the other's nest or young

Recordings in audio recording below aren't gap free - the whole event lasted around 30 minutes in total, I've just shortened somewhat.

1: Magpie alarm call
2: Magpie alarm call - sounds more upset
3: Tawny Owl calls interspersed with the Magpie alarm calls
4: Tawny Owl calls interspersed with the Magpie alarm calls
5: Calmer - can hear both a Tawny Owl and Magpie
6: Very upset sounding Magpie
7: Sounds quite nasty - possible physical interaction or faceoff between Magpie and Tawny Owl ? Can hear both in recording
8: Finishes off with 5 minutes or so of solitary Magpie Calls (just one clip here)

I did go outside with a torch to check to see if any injured birds (either from a fight between themselves or a predator) but nothing visible and was all quiet by then

Any ideas on the squabbling in part 7 ? I was wondering it this is actually a physical interaction between Magpie/s and Tawny Owl/s - sounds quite nasty but could just be vocal

as an aside still not managed to work out this "Howl"- can't find a match with any Mammal.

My daughter though thought she re-heard a few nights back (she had her windows open) - BirdNET unfortunately didn't record as didn't pick up a bird at the time, so put camera on the following night and to my surprise we had a Badger visiting literally feet away from the Kitchen window - not the mammal making the noise but was well pleased to see on the Wildlife camera I had no idea there were any even in the area. I will continue to investigate to find out what made the original Howl. I still think either a Fox with a slight odd vocalisation or a Deer - but can't find any near matches to either online.



Had a nice recording of one of the local Buzzards the other day - must have been calling as it flew overhead as nice clear recording:

 

[LittleBitOfBreadNoCheese]

now up to 185k bird recordings with the BirdNET Pi

this was an interesting set of recordings from a few weeks back - be interested in people's thoughts

Woke up in the night and decided to look at the BirdNET Pi to see if any owls had been detected and saw over 50 instances of Magpie around 00:30 (which seemed odd - very late) interspersed with Tawny Owl - so thought I'd have a detailed listen to see what was going on.
Can't work out if group (or individuals) disturbed the other, territorial, both the Magpie/s and Tawny Owls were disturbed by a common foe (Fox, Cat etc) or one of the sets of the birds attacked the other's nest or young

Recordings in audio recording below aren't gap free - the whole event lasted around 30 minutes in total, I've just shortened somewhat.

1: Magpie alarm call
2: Magpie alarm call - sounds more upset
3: Tawny Owl calls interspersed with the Magpie alarm calls
4: Tawny Owl calls interspersed with the Magpie alarm calls
5: Calmer - can hear both a Tawny Owl and Magpie
6: Very upset sounding Magpie
7: Sounds quite nasty - possible physical interaction or faceoff between Magpie and Tawny Owl ? Can hear both in recording
8: Finishes off with 5 minutes or so of solitary Magpie Calls (just one clip here)

I did go outside with a torch to check to see if any injured birds (either from a fight between themselves or a predator) but nothing visible and was all quiet by then

Any ideas on the squabbling in part 7 ? I was wondering it this is actually a physical interaction between Magpie/s and Tawny Owl/s - sounds quite nasty but could just be vocal

as an aside still not managed to work out this "Howl"- can't find a match with any Mammal.

My daughter though thought she re-heard a few nights back (she had her windows open) - BirdNET unfortunately didn't record as didn't pick up a bird at the time, so put camera on the following night and to my surprise we had a Badger visiting literally feet away from the Kitchen window - not the mammal making the noise but was well pleased to see on the Wildlife camera I had no idea there were any even in the area. I will continue to investigate to find out what made the original Howl. I still think either a Fox with a slight odd vocalisation or a Deer - but can't find any near matches to either online.

View attachment 1586770

Had a nice recording of one of the local Buzzards the other day - must have been calling as it flew overhead as nice clear recording:

I've had a near identical siutation two nights in a row. All day time birdsong had finished, even the Blackbird, when suddenly a Magpie kicked off and a Tawny Owl called in response. Both were out of sight so I have no idea what was actually happening. However the time was just on nightfall and past normal Magpie bedtime so I figure most likely the Tawny was getting active.

 

[alexbelgium]

very interesting thankyou for heads-up on that report

ref BirdNET PI - is it possible to change the recordings to 24 bit please ? I tried but failed

I tried changing all references in birdnet_recording script from s_16LE to s_24LE to no avail - just get very over saturated outputs

Hi, actually I have tried it on the add-on for homeassistant that I have developed. It allows to have birdnet pi on any x64 or aarch64 device running HA. A positive consequence is that it allows to run two instances of BirdNET pi on the same hardware, with the same sound input, and check the impact of a single parameter change.

Currently, birdnet pi transforms all input in a mono 16 bits 48khz file which is how the model was trained. I tested 24 bits with a simple sed change here : Add 24BITS_ANALYSIS · alexbelgium/hassio-addons@f443351

In the end it made no differences that I could see after 1 day on the two systems. I guess that as the model is trained on 16bits it doesn't impact

 

[Buckster]

alexbelgium thanks very much for the reply and apologises for the delay in answering - will take a look at your code thankyou

BirdNET PI in garden still being brilliant its up to 256k recordings now !

Was VERY pleased with this recording the other day - Peregrine and really clear. I'm not sure what was going on as didn't see (was out) - not sure if was having an argument, was "talking" to another Peregrine or had caught something

so far I've recordings Raptor and Owl wise of:

Buzzard, Red Kite, Marsh Harrier, Sparrowhawk, Kestrel, Peregrine, Merlin (possible)
Barn Owl, Tawny Owl, Little Owl

impressive device

 

[Buckster]

Noticeable up-tick in overflying waders at night over the last week - BirdNET has picked up in that period:

Grey Plover
Redshank
Spotted Redshank
Little Ringed Plover
Ringed Plover
Common Sandpiper
Green Sandpiper