How do you review/identify audio observations?

[ken_ohio]

I haven’t done a lot of identifying in iNat mostly because I only know birds well enough to do it. And, those go REALLY fast. But, lately I have been trying my hand… uh… ear at audio observations. I don’t know why I didn’t think of doing that sooner! I can bird by ear pretty well.
(I have been learning dragonflies and damselflies. But, I only know the common species.)

Here’s how I identify audio observations:

1. Filter for only Birds and Unknown.
2. Select a Place in the states/provinces/counties in and around Ohio.
3. Go to it!

Now, things I have noticed:

A lot of people upload Gray Tree Frog recordings and don’t have a clue about what it is. That’s cool! They are very loud and very difficult to find even during the day. I have only found two or three while hiking over the years only by luck. I will sometimes add a link to this website I like and add a comment to their observation that it’s a good website.

https://musicofnature.com/calls-of-frogs-and-toads-of-the-northeast/

There are a lot of audio files that I can’t hear anything. And, I hate to say it but, I look at that person’s ID and skip the rest of theirs by marking them as Reviewed. I should add that I use public library computers so, I can’t install any add-ons to the browser to increase the sound past 100%. If anyone has suggestions about how I can increase the volume of a recording, please let me know.

I also do the opposite: If I find a person that has a good recording (quality sound and loud enough), I click on their icon and search their observations for more audio observations that need identifying.

If a bird call/song is difficult to hear, I sometimes put where I heard it in the comments. Example: “Started at the 12 second mark”. Sometimes, I see identifiers just add a comment that they can’t hear anything. So, I try to point out where I heard it.

I have used the Merlin app to make recordings myself. I have found that I need to increase the sound recording - which I do using Audacity (portable version off a flash drive). I also found this really cool website created by an iNat user. It was posted in another Forum topic. It uses Audacity to clean up a recording. It works well. I used it before learning how to use Audacity myself.

https://forum.inaturalist.org/t/quick-and-easy-tool-for-cleaning-up-audio-observations/65427/10

https://normalizer.pages.dev/

If you have any tips on how to be a better audio identifier, please add them here.

[lappelbaum]

@ken_ohio

I haven’t done a lot of identifying in iNat mostly because I only know birds well enough to do it. And, those go REALLY fast.

Not really, unless you are limiting yourself to only IDing birds in Ohio.
Here’s a Northern Cardinal in Virginia uploaded in 2019 that needed an ID until I added one just now: https://www.inaturalist.org/observations/27997333

I’m more likely to skip over obs with a lot of background noise (people talking, cars, etc) than ones that are really quiet. I wouldn’t skip over a person’s entire collection based on one recording. Maybe that bird happened to be really far away, but their other ones are good quality recordings. I’ll keep my recording on the quieter side if increasing the volume too much with Audacity causes distortion.

[ken_ohio]

Yes. I only look to identify birds in Ohio and the surrounding areas. I’ll try expanding the area. I wish we could define an area on a map for identification like on the Explore map.

I have found that people are pretty consistent with what they post as far as audio files. I just wish I could easily make them louder. I was thinking of bringing my over-the-ear headphones into the library. They are better than my ear buds - which are pretty good.

I also find that there are recordings with a lot of noise - traffic and people talking. One recording just seemed to be wind chimes.

[lappelbaum]

@ken_ohio

I wish we could define an area on a map for identification like on the Explore map.

You can. Draw your circle or rectangle on the map on the Explore page then in Filters click on Identify in the bottom right of the pop-up box. Also there are existing places that are larger than a state, such as Eastern United States. I usually use the place American Southeast when I search by “Birds”, but I don’t limit to only certain areas when doing IDs of some species. I know what a Blue Jay, Northern Cardinal, and Carolina Wren sound like where ever they are.

[gcwarbler]

@ken_ohio

I wish we could define an area on a map for identification like on the Explore map.

Am I missing something? On my Mac, the Identify page has a “Place” field in which I can narrow my search for a set of observations to anywhere that is a Place in iNat’s geography (e.g. my local park, or even my own home project kml-defined area). Thus you can subset your Identify page by any county, state, or country if you so desire. Maybe you’re thinking of even more fine-grained or customized areas; that would be nice as well.

[phma]

I don’t filter for audio observations. I filter for anything that still needs ID and was uploaded the previous day, and sometimes there are audio observations.

I can usually identify a coquí (which I’ve heard in the evening when visiting Puerto Rico, but the obs I find are more often in Hawaii), a whippoorwill, and a chuck-will’s-widow (both local), and I’ve identified sounds as passerine, owl (checking the time of observation helps with this), bird, and mammal.

[joedziewa]

Spectrograms can be useful. If you can run Merlin on your phone and its Android, there’s a pretty versatile open source app that should be able to work with whatever sound the mic can pick up. It’s on F-Droid and called Audio Spectrum Analyzer. It has worked well for me with the phone mic at ear buds, headphones, or speaker. It’s been great for me with finding exact frequencies or pulse rates, which is helpful in a lot of IDs(like gray treefrog complex) and can be difficult to know for sure by ear. If there’s any frequencies that you can’t hear well or there is a lot of overlapping, headphones to the mic may help find and identify the sound too.

If you’re delving into unknowns, insects might be good to learn too. There will be a lot soon. Common true katydids will be taking the spotlight over from the gray treefrogs. They and northern mole crickets will also be showing up in Frogs and Toads. A good starting point for our singing insects is http://songsofinsects.com . It has a nice thumbnail guide. For more comprehensive info on crickets and katydids in N American north of Mexico, https://orthsoc.org/sina/index.htm is great.

You didn’t mention, so just to be sure, “has sounds” is a good box to check in filters :)

[hagrimes]

I find it sooo much easier to ID bird audio than blurry photos of birds flying away! When I’m struggling to hear, I just look at the spectrogram. I find my eyes are a lot more accurate than my ears in many cases. The super low-tech way to approach it on a pinch is to simply play the audio with the bottom mic on your phone resting against the speaker/headphones and have Merlin record it so it generates a spectrogram. Merlin won’t necessarily give away the answer, but you the human can at least take a closer look at it and see what’s going on in the recording.

I love the bird-frogs, those are my favorites.

[bcs_mass]

I just want to say “thank you” to the audio identifiers. There was one audio observation of mine that turned out to be a green frog - and I’d have never known what to look for, as they sound so different from other local frogs. Thank you.

[ken_ohio]

I have used the Place field. I would like states around Ohio. There is “Midwest” and “Lower Midwest”. I can see where these are defines in the Places page. They don’t include states to the east or south of Ohio. I don’t want to define my own Places just for identification to include states to the south and east of Ohio. The iNat Help says to think about making a lot of Places.

[ken_ohio]

That is exactly what I was looking for. It’s not the Identify web page with all it’s little options/behaviors. But, it’s still getting me the geographical area I would like.

[ken_ohio]

Look at the answer just above yours by lappelbaum.

[ken_ohio]

Thanks for those links.

I use this webpage: https://musicofnature.com/calls-of-frogs-and-toads-of-the-northeast/

[lappelbaum]

@ken_ohio

It’s not the Identify web page with all it’s little options/behaviors.

Click where the blue arrow is pointing to get to Identify web page with area defined by your circle

[lappelbaum]

Here I had my cursor hovering over Identify (blue circle) and you can see that the link leads to Identify page (URL at bottom of page) with same location parameters (yellow underline) as the Explore page (current URL at top).

[ken_ohio]

Thanks, again!

I do like that Identify page.

[sandboa]

I do a lot of audio identification, but mostly frogs and toads. I also post a lot of audio files and getting them identified by others can be a frustrating task.

It can also be frustrating that so many recordings are of such poor quality that they can be very hard to hear. I often end up downloading their recording, opening them in Audacity, cleaning them up enough to hear the target and then identifying. It takes a bit of time unfortunately, but it removes a lot of recording from the “Needs ID” state.

I don’t search bird recordings for misidentified frogs, but if someone tags me on one I will take a listen. I do find a lot of insects labeled as frogs (the Northern Mole Cricket is a very frequent ID error, but they do sound like toads!).
I also listen to recordings of Giant Toad vs. Lesser Nighthawk in areas where their ranges overlap since they have very similar calls and I’ve found a few that were probably misidentifications (https://frogcalls.blogspot.com/2015/01/cane-toad-vs-lesser-nighthawk.html).

I have a blog that has a lot of frog species recordings explained - https://frogcalls.blogspot.com - where you can find the calls of around 100 US species plus some from other places around the world.

One possibly useful (?) comparison is the one that compares the calls of the two gray treefrog species - https://frogcalls.blogspot.com/2016/03/gray-treefrogs-hyla-versicolor-vs-hyla_29.html for those trying to tell those two apart.

For Australian frog species, the Australian Museum’s Frog ID page is a fantastic resource for sorting species by state and then hearing recordings of the species found there - https://www.frogid.net.au/frogs

[bat_dude]

Bat calls. I have been reviewing posted bat calls when time permits AND the actual call files are included. Many Citizen scientists are simply posting screen shots that are never diagnostic for ID w/o the call file and therefore can virtually never be confidently ID’d as the display parameters are also usually not in a standard format. Most if not all could be valuable distribution records even for common and well known species if those posting includes the actual call file. A key issue with bat calls relates to the software used to review the files for the characteristic vocal signatures of each bat species. As a pioneer in using and developing acoustic methods to ID free flying bats since the mid 1990s and taught > 30 workshops a key take home message is NOT EVERY call file can be ID’d to a specific species.

iNat still needs to allow posting of ZC format call files for bats as this is a valid format. Not unlike images posted as a PNG vs JPG image file. Just different formats to achieve the same thing, convey important information on the taxa presented for valid identification. I would say an important distinction on bat calls is there are very few who are able to “confirm” or have time to confirm IDs to upgrade a posted observation to “Research grade”. So if I agree or disagree with a posted ID and correct the IDs that still only results in a single verification and limits the upgrading of observations to Research Grade. I have been storing ALL Neotropical call files in my master call database (> 2 million records) and each and every record becomes a valid distribution point for the confirmed species and contributes to a better understanding of species and habitat distribution. I have yet to be able to reach the “Powers that be” to get teh change to allow Zero Crossings file format to be added for bat calls. This would be a HUGE contribution to knowledge of bat distributions.
So my comments on this issue.

[ken_ohio]

I turned this into a feature request. It was moved to the General topic area because I didn’t have a good solution. But, I wanted more users to know about the Explore map / circle tool way of defining a geographical area for identification. I think it is better. But, I also understand that the other method of using Places might work better for someone else.

If you want to take a look and comment, it is under the title:

“A better way to define a geographical area for Identifications”