Ryan McClarence

An enjoyable couple of hours at South Gare today, the Olive-backed Pipit being the main target as it was a lifer and it showed well at times. My first (6) Snow bunting of the year were also good to see and made the trip even more worthwhile. #phonescoped

My first lifer of 2025, the Eastern yellow wagtail at Winteringham, Lincolnshire. It was very blustery and was quite mobile feeding with 10+ Pied wagtail, but showed really close at times. #phonescoped

Sand martin just appeared in front of Ness hide at Far Ings nature reserve. @lincsbirding.bsky.social

Soundscape January 14, 2025 04:07 @ Norton Common

A nice clear Barn owl detection this morning. The detection was rated at 99% confidence with Barn owl probability of 5% app.birdweather.com/soundscapes/...

Still running probability at 1.5%, with a confidence of 63%, still throwing up a few errors, but they're easy enough to weed out.

#BWPUC

You can get to them on mobile, but it seems to be more difficult than it needs to be!

Marsh harrier

An enjoyable 5.2 mile walk out on patch rewarded me with a decent return of 42 species.
The highlights were Little grebe and Marsh harrier as I didn't manage to record either in the whole of 2024! #patchbirding

ebird.org/checklist/S2...

I had P at the lowest on 0.5. Currently, I'm upped it to 1.5, but any higher, and it doesn't pick any Owls up. I've got C at 63, and I think it's the higher confidence rather than a higher probability that is giving me the improvement.

The vast majority of my PUC detections are coming from the local Robin. I'm considering turning off detections for Robin to see if that further reduces the errors. However, over the past few days, I've started to up probability and confidence and have already noticed an improvement.

On patch, I managed only 91 species on 52 checklists, however in a usual year I tend to submit between 70-100. Still, 3 new species got on the patchlist with 2 flyover Ring-necked parakeet on 18th Sept, and the flood in May enticed 2 Shelduck down on the 2nd and a Redshank on the 7th. #patchbirding

Snow geese. 13th November. Swine Moor. Collared flycatcher. 3rd May. Spurn. Storm petrel. 1st September. Spurn. Black-winged pratincole. 10th July. Finningley, South Yorkshire.

My Yorkshire list crept up to 339. New additions being Collared flycatcher at Spurn, Black-winged pratincole when it eventually moved over the border, finally got Storm petrel, on a Spurn seawatch, Pale-legged leaf warbler at Bempton and ending with the Snow geese at Swine Moor. #YorkshireBirding

Whiskered tern. 8th June. Woodhall Spa Airfield, Lincolnshire. Dark-eyed junco. 26th April. Hartlepool, Cleveland. Cirl bunting. 15th August. Labrador Bay, Devon. Black-winged pratincole. 29th June. Finningley, in both Nottinghamshire and Yorkshire. Cory's and Great shearwater. 17th August, at sea on the Scillonian crossing. Collared flycatcher. 3rd May. Spurn, Yorkshire.

My birding in 2024 both started and ended with a mega. The Essex Northern waterthrush and the Kent Yellow warbler. Other lifers: BW pratincole, Cirl bunting, Collared flycatcher, Cory's & Great shearwater, Dark-eyed junco, Pale-legged leaf warbler, Red-breasted goose and Whiskered tern. #UKBirding

That may come in a future update, but for now, this is the best way to deal with the definate not happening species. bsky.app/profile/donn...
One feature I would really like, is to set confidence levels to individual species.

As you say, Steve, it's a new product, and with what it's doing, it's bound to get things wrong. Some owners just set it and leave it, but I try to keep my PUC as accurate as possible and actively check and delete errors every day, and for my own personal records, I think the PUC is brilliant

Absolutely, it will. The AI will always be learning and improving. I've only had my PUC a few weeks, but it's made checking a nights Nocmig so much easier and faster. Any records that seem dodgy, I check, and if they're wrong, just delete them.

I've got a PUC. It's far from perfect, and it throws out plenty of weird suggestions, but it does get plenty correct too. I've listened to quite a few of the other PUCs recordings. One had Great Bittern, and I'm sure it was somebody snoring 😆

Where have you seen about data being fed into eBird? I understood the only use of eBird is to set the probability of the species, so it's coming from rather than going to?
I would also like to think anything put on eBird from a PUC owner should be backed up with a recording?

Happy to help Steve. Mine took about an 1hr 30 to upload 2500 detections. There's no prior warning the PUC shuts down after uploading, the more tips we share between us the better.
Talking of detections, this is pretty cool. Just filter to find your PUC.
app.birdweather.com/detections/ #BWPUC

I am still doing ok with 0.1% P and 44% C. I like the lower P, so it will show anything and at least give me the chance to double-check it. Anything nearly impossible, I hide detections, like I've done with Crested tit. It's the local Starlings though, that are causing me the most mis IDs!

BirdWeather - a living library of bird vocalizations. Join a growing community of citizen scientists with PUC, our AI powered bioacoustics platform.

Portable Universe Codec.
www.birdweather.com

It's a listening device, John, I've got it running in my garden 24/7, and it records bird calls and song and IDs them. Some are incorrect but you can easily listen to and delete them. It's so much easier than using a recorder and mic for nocmig.

If the PUC is powered but offline, it will still detect. When you get it back online, it will upload all detections, but once it's finished, the PUC will shutdown, so it will need to be powered on again, which I didn't find out until this morning so have missed a full nights recording! #BWPUC (2/2)

Thought I'd share a problem I had with my PUC and how I sorted it. It went offline and didn't go back online with a power cycle. Turns out we'd had a power cut and didn't reconnect back to the WiFi. To do this, go into settings in the app, then setup my PUC and follow the instructions. #BWPUC (1/2)

Another tip to new PUC users. If you have lowered your probability down to 0.1%, you will get a couple of species suggested that are extremely unlikely for your location. Click on detections of that species, then probability, then change the Birdnet default to no longer show them. #BWPUC

A murky but productive walk on patch, I took the thermal to look (unsuccessfully) for Jack snipe, but had a decent count of 31 Common. Then 2 very vocal Raven (only 3rd patch record) came in from the East. They landed briefly in the distance, then continued West.
ebird.org/checklist/S2...
#UKBirding

I haven't, I'm happy with the results I'm getting now without one, but it would be interesting to see if one would help.

They were something I had considered, but I ended up making a housing for the PUC instead, which seems to be keeping any wind or rain from directly hitting it.
I'm tempted to try and make some form of parabolic dish to add to the front of it to try and help it collect as much sound as possible.

BirdWeather Community Welcome! This forum is a place to discuss all things BirdWeather.

community.birdweather.com/categories

No problem. You should be able to get more info here or get an answer to any questions you may have.
Delivered the puc cost me £230.57.

I've had my PUC 2 weeks now, and I've really enjoyed using it. It's not perfect yet, but it'll only get better with future firmware updates and the improvement of the ID software. I've no doubt in the busier months it'll be brilliant for nocmig. The only negative is the price! #BWPUC

By dropping down to 35%, has it correctly picked up any new species for you with a C under 40%? I'm tempted to lower mine to make sure I don't miss anything. With my current settings instead of default, my puc has detected Barn owl, Green sandpiper, and Tawny owl. #bwpuc

What probability (P) and confidence (C) have people got their PUC set at? I've been playing around and settled on 0.1% P and upped C to 55%, and it seems to be going OK. With the lower P, I'm obviously getting more wrong IDs, but I've gained a few correct IDs the default would have missed. #BWPUC

Regarding confidence, that is again a bit hit and miss, especially when it is a non bird sound. The erroneous Coot and Moorhen detections I had been getting had a confidence of at least 75%, despite it being the handle on my gate creaking!