> I’ve never seen Merlin’s Sound ID produce an obviously-wrong identification
Well, good luck identifying mockingbirds, lol. But that's a tricky case.
Anyway, it's a nice way to use AI, unlike so many others that pop up lately.
I wish they'd open source it all though, especially if they are getting help from volunteers.
joshuahedlund•6mo ago
As a heavy user of Merlin, it definitely isn’t perfect - especially with all the mimic birds - to the point that there are complaints about beginners polluting the citizen science database with erroneous IDs from Merlin.
But it’s very very good.
shmerl•6mo ago
Yeah, it's generally pretty accurate.
AnotherGoodName•6mo ago
Merlin generally gets Mockingbirds correct which is part of it's magic.
Mockingbirds do a minimum of three repeats of any particular sound before moving onto the next. So the songs are done more times in a row than the real species would and then immediately it turns into a Robin (or something else). I'm sure there's subtle audio differences between a mockingbird and the real thing too that Merlin can detect.
A feature I'd love would be if Merlin told you the call the bird was making rather than just the bird ID. eg. "Blue Jay mimicking a Red Shouldered Hawk" would be a cool bit of info it could give since it really does catch the mimics out really well. It currently just says "Blue Jay" when it hears a blue jay trying to be a red shouldered hawk.
shmerl•6mo ago
It does make mistakes with mockingbirds in practice.
joshuahedlund•6mo ago
I have stared at a Blue Jay mimicking while Merlin repeatedly labeled it Red-Shouldered Hawk. I’ve seen it pop up a bunch of suspicious one-offs around a mockingbird as well (same with Gray Catbird, another mimic).
But I agree with you that it gets those things correct most of the time and it also seems to be improving over time.
thaumasiotes•6mo ago
Seek by iNaturalist has the same problem: if you wiggle your camera around enough, it will give you a species identification. But that species identification will be spurious. It's not reliable at all.
Pointing this out on HN has sometimes resulted in a lot of upvotes and sometimes in a lot of downvotes. I don't know why. In all cases, Seek identifications are unlikely to be correct.
(I also got one response saying I was wrong to try to tar iNaturalist by association with the unaffiliated app "Seek by iNaturalist". As the name of the app suggests, they are not in fact unaffiliated.)
femto•6mo ago
> Sometimes magic is just someone spending more time on something...
I'd suggest the real metric is "attention to detail", not "time spent". It just so happens that attention to detail often requires time. The real magic happens if one can manage to take care of the details without blowing out the time.
shmerl•6mo ago
Well, good luck identifying mockingbirds, lol. But that's a tricky case.
Anyway, it's a nice way to use AI, unlike so many others that pop up lately.
I wish they'd open source it all though, especially if they are getting help from volunteers.
joshuahedlund•6mo ago
But it’s very very good.
shmerl•6mo ago
AnotherGoodName•6mo ago
Mockingbirds do a minimum of three repeats of any particular sound before moving onto the next. So the songs are done more times in a row than the real species would and then immediately it turns into a Robin (or something else). I'm sure there's subtle audio differences between a mockingbird and the real thing too that Merlin can detect.
A feature I'd love would be if Merlin told you the call the bird was making rather than just the bird ID. eg. "Blue Jay mimicking a Red Shouldered Hawk" would be a cool bit of info it could give since it really does catch the mimics out really well. It currently just says "Blue Jay" when it hears a blue jay trying to be a red shouldered hawk.
shmerl•6mo ago
joshuahedlund•6mo ago
But I agree with you that it gets those things correct most of the time and it also seems to be improving over time.
thaumasiotes•6mo ago
Pointing this out on HN has sometimes resulted in a lot of upvotes and sometimes in a lot of downvotes. I don't know why. In all cases, Seek identifications are unlikely to be correct.
(I also got one response saying I was wrong to try to tar iNaturalist by association with the unaffiliated app "Seek by iNaturalist". As the name of the app suggests, they are not in fact unaffiliated.)