In the age of the Merlin Bird ID app, it is very easy for observers to directly upload a lengthy audio file that supposedly contains a certain bird species. In many cases, observers will upload the same minutes-long recording multiple times to create an observation for each of the species present (or each of the species suggested by Merlin).
As a frequent identifier of bird sounds on iNat, this is excruciating. It is common for the indicated species to not actually be in the recording. Or, with no indication of when the indicated species occurs, it can be difficult to pick out a single call note in the background of a 60 second audio recording. As a result, itās clear that some identifiers just skip past these recordings. Other times, someone disagrees with the observerās ID simply because the first or loudest bird is something else entirely.
It would be great to have some sort of warning pop up when someone attempts to upload a 30+ second recording that suggests they either (a) trim the recording to focus on the species indicated in the observation or (b) include a time stamp in the Notes (Merlin makes it easy to find the time stamps for a suggested species, even if you donāt know the ID).
How do you handle these? Are you too tempted to skip these longer recordings, especially when you see itās been uploaded 6 times for different species?
(Iām not posting this in the Feature Request forum for now, as Iād love to hear more about what other folks think about this problem and how they approach it.)
I havenāt noticed too many long recordings. What do you consider too long? I do upload the occasional Merlin recording but I always keep them fairly short (~30 secs). I think I may have a couple longer than that but the bird was doing something cool in those. Also, I only upload ones where the subject bird is calling in the first 10 secs of the recording.
Lately, Iāve seen quite a few in the 2ā3 min range where the species indicated only calls once or twice. Personally, once a recording gets past 30 seconds, Iām less likely to listen through the whole thing if the bird doesnāt call in the first few seconds.
While I donāt usually bother to post audio recordings of birds Iāve been able to take pictures of, If I can hear one using merlin but canāt see it and its loud/clear enough to reliably identify I typically add a screenshot of the spectrogram for that birdās call along with the audio
For what itās worth, I donāt consider an audio observation to be ālongā until the total length of time is over two minutes. Of course Iād prefer a cropped audio clip, but many users wonāt know what is or isnāt relevant ā or, they may not know how to edit audio. If it became an issue in my region Iād politely reach out to the user(s) who cause the problem.
Iām much more frustrated with recordings that havenāt normalized the sound, because otherwise I canāt hear the bird on my laptopās speakers, and itās a pain to get out headphones.
That said, no, Iām not interested in playing whereās-the-bird-youāre-asking-about, whether itās in an uncropped photo or an uncropped audio file. I do enough of that with my own bad photos and recordingsā¦
Sometimes they include screenshots! But, nowadays, nearly everyone in North America who is recording bird sounds on their phone is using Merlin, so itās a pretty safe guess. (Of course, if they arenāt using Merlin, then I assume they know which sound on the recording is the bird they are identifying, and see above.)
Iāve assumed most are using the iNat app to record the bird songs. Thatās what Iāve done. And if the bird hasnāt given a decent vocalization within 30 seconds I discard that recording and try again.
I came across several 5 minute recordings a year or so ago. I asked the uploader to say where in the recording the sounds were that they were interested in, because 5 minutes was too long to expect anyone to listen to. They said it was part of some frog census, and they had to do 5 minute recordings. From what I saw none of their observations were given IDs beyond āFrogs and Toadsā. Whatever the recordings were for, iNat wasnāt the place for them.
The thing that made me switch is how useful it is to see the spectrogram on the screen, very nearly in real time. There are some bird sounds that I struggle to distinguish by ear, but seeing the shape on the screen helps me to hear the difference.
https://www.inaturalist.org/observations/344859671
This is almost three minutes long; I heard a bird and walked toward it, with the recorder running, in case I frightened it away. I provided timestamps when someone asked, but it still hasnāt been identified.
Most of my recordings are shorter than a minute, but sometimes I hear a bird and start the recorder, but the bird doesnāt say anything for a minute.
I donāt have Merlin. I do have Audacity, which has helped me to measure the pitch of a bird, but it doesnāt tell me what kind of bird it is.
With my own recordings I consider a āshortā recording less than 10-15 seconds (nocturnal flight call recordings often get cropped to only a couple seconds) and get bored of listening if they go beyond 25-30 seconds. I listen to bird recordings regularly for work, and 3 minutes feels like a very long time to be sitting inside and listening to a single soundscape. For the purposes of iNaturalist I feel like that long is excessive and multiple <30 second recordings would be preferred by anyone. Even in 30 seconds there can often be enough species vocalizing that itās unclear which is the target.
But on the other hand, it does help a lot to hear repetitions of a vocalization because it often happens that a single vocalization is unusual and not representative of what the species (or the individual) normally sounds like.
The tool āiNaturalist Sound Classifierā with āBirdNet v 2.4ā suggested that this could be Melospiza melodia (90.0 to 100.0s) or Parus major (57.0 - 60.0s). āPerch V2ā also suggests Melospiza melodia
I am not a bird person, but I will try and get audios. I have trouble processing sound and I often donāt know which bird sounds I have ācapturedā. If my husband has the time I ask him to tell me which is which so that I can put it in the notes but often I have to rely on the iNats community.
I hadnāt heard of Merlin and was unaware being able to edit the audio once uploaded - all my bird recordings are done through the iNats app.
A frog researcher told me he needs recordings longer than 10 secs (sth Aust)
