Which bird sound to identify in a recording of many birds

Is there a protocol for deciding which bird sound to identify when a recording has multiple bird sounds?

I believe the person posting the recording should specify which song or call, if possible, along the same lines as specifying which plant or insect is the focus of a photograph.

6 Likes

Thanks @Thunderhead I’ve encountered many recordings where that is not specified and wondered whether there were some ideas about how to deal with those.

1 Like

If there’s clearly a bird sound in the foreground I’ll go with that, but if it’s otherwise unstated I’ll leave a comment asking which they refer to and then stating each species in the recording, either specifying by sound or time.

7 Likes

Thanks @natemarchessault I didn’t thinking of specifying the time when noting several birds.

1 Like

I’ve had identifiers specify other calls even when I’ve indicated (by time stamps) the one I want identified. That’s how I ‘found’ a lifer that my brain didn’t even notice along with a louder call I did notice and wanted identified.

People who identify other things (I might have missed) in recordings or photos are heroes!

8 Likes

After I posted that I remembered I do that too but failed to mention it. All of this does nothing for the “data” but I think people enjoy learning about the birds they were hearing and which were in the recording even if they’re not birders or have a specific interest in birds, and to me there’s value in that.

Edit: Not trying to claim myself as a hero!!! Just trying to say doing something purely for the fact that someone else may enjoy it is a nice thing to do, in the hopes others may follow suit!

3 Likes

Oh, yes! I think that is the best of iNaturalist.:hatching_chick: learning and sharing knowledge.

3 Likes

Coincidentally, I just had this ~exact~ situation pop up this week. I noted the time stamp for the “bird?” sound I intended in the Comments, but there were also other prominent bird sounds. This was in part because we cannot edit or trim a sound file in the iOS app. So, I made a Voice Memo recording that focused more on the squeal.

@yubabirder and I had a conversation about it, and now we are not even sure the squeal is a bird (although, I do not know what else it could be).

https://www.inaturalist.org/observations/109074118

  1. I listen out for the more prominent call and id that, assuming that the observer has not mentioned the species already.
  2. In case the observer has mentioned the species then often it is corroborative however if there are other calls I mention the time and also tell them to duplicate the post and upload it as the other species(s) clearly mentioning the time
  3. I often ask the observer what they are identifying

Some of the calls I id are very poorly recorded / edited so I download them and edit them on either audacity or ocenaudio and try and see what was the target sound - again assuming the mic was pointed in the direction of the sound - something which I am noticing is sadly not the case at all.

2 Likes

For editing I would think it would be very hit or miss and situational, although I applaud your efforts.

Pointing the microphone in a certain direction will help but I would imagine most record with their cell phone which is somewhat limited. This would increase the volume of what the mic is pointed at versus everything else. In post though, one is unable to isolate sounds based on where a mic is pointed, other than playing with the volume of the left and right channels (phone mics probably record mono so it would be the same for both channels). If it is quiet, one could increase the volume but it would do so for the whole recording.

Reason being is that an audio recording basically captures a range of frequencies which is what we are limited to being able to manipulate. Think of it like a range of sounds from high to low pitch that we are able to increase/decrease the volume of each. If a call is high or low frequency relative to other background noise we can reduce those frequencies in order to isolate the target sound. Conversely, if the unwanted noises overlap in frequency with the target sound, there is not much one can do. The spectrogram view on audacity can be helpful for seeing what frequencies these calls are at and be used to filter out unwanted frequencies.

Anyway, this might all be stuff you know but I thought I would share in case it is helpful to you. In my opinion, if you listen to a sound here and it is still too quiet when you’re at max volume that would be one situation you could export it to audacity, and if the sound is significantly higher or lower in pitch than the background noise that would be another. Otherwise, you may want to spare yourself the time and effort and chalk it up to a lost cause :slightly_smiling_face:. More time to ID other recordings!

1 Like

Thanks for all these replies. I will take them all into account when I attempt to id bird sounds. I am a beginner but will be trying to id more, to increase my knowledge.

1 Like

editing poor recordings do not always work. But it does in, atleast, some of the cases.

Usually I end up recommending that the observer use a better recording app , like recforge II (the free version is great). And I also send them a link on how to improve their sound recordings.

I wish that inaturalist generated spectrograms from the audio files directly (like ebird) but I think that is a feature that will take a while to come around.

1 Like

I have no idea what’s required to implement something like that but I agree it would be really cool. I’m consistently impressed by the quality of eBird’s auto-generated spectrograms, I’m never able to make them as good using Audacity or Raven Lite

I agree,

I do find that ocenaudio Audio Screenshot is a reasonable “spectrogram” although static,

I post them alongside the audio files (rather as the first image — but nothing as nice and effective as a live spectrogram that seems to speak to the reviewer / observer

Some examples are here

  1. Mountain Scops Owl
  2. White-bellied Flying Squirrel
  3. Domestic Dog
1 Like