No warning displayed when audio files are too large, and they block uploading other observations

Seems like the iNaturalist app is stuck on uploading the audio files for this observation, which is currently blocking uploading of all my unuploaded observations. There are 5 audio files: 2 approx 7 minutes in length and 3 about 2 minutes in length.

This is the observation with the issue:

You could post an edited portion of the soundfile, and then link to the main files stored external to iNat. It’s hard to imagine 20 minutes of audio being required to evidence an observation, but I can understand there might be other reasons for the length.

1 Like

I have this issue all the time with recording audio. It’s the downside of recording high-quality or long audio files when uploading to iNat, and it’s the reason I only put my bird audio on eBird. For my long insect recordings, I will usually chop up the recording into several pieces and leave a note that I had to do so due to iNaturalist file size constraints.

I would nog upload more than 20MB per upload in total

There is a maximum upload limit of 20 Mb per upload. If you have several files for one observation but they total more than 20 Mb, it will bomb out. You then upload one file and save the ob, upload the next image/audio to it etc. If a single file is larger than 20 Mb, it will not upload. You will have to make it smaller.


Obviously not great for UIX!

Seems like it would be preferable to increase the maximum observation size limit of each file and all files? I suppose iNat doesn’t want to store too much on their server?

Other alternatives (some mentioned) include:

  • upload in app to another file storage service, e.g. cloud providers like Mega, or distributed file storage like IPFS (e.g. this service,, and post a link. Ways to search for and retrieve those links within the app would help, rather than being “lost” / not searchable in notes/comments.
  • find some way to sustainably pay for storage on iNat servers (a hard problem)
  • forcing users to use another provider like eBird that doesn’t have an audio limit, or split files up into chunks, isn’t ideal.
  • have better interoperability / bridges between other sites like xeno-canto and eBird. This is hard and complex, but seems like a worthwhile approach, but also broadens the scope of this issue greatly.
1 Like

I like that you summarise the discussion/solutions!

For me, I think the key thing is that iNat isn’t specifically for documenting the full range of voice for a species, or in the context of image/video the full range of behaviours of a taxa. It is more about documenting the moment of contact. Any image/video/audio is primarily about evidencing the encounter/presence of the taxa rather than studying it in depth. Even in a photo, there is a limit to the resolution that iNat will store it with.

We sometimes see observations with 20 or more photos, and in many of those cases there might be enough evidence within each one so the argument can be made for only needing the one photo. However, each observer also has other reasons they might be uploading that observation, such as viewing that animal from all angles to show markings so that they might recognise it in future as the same animal in a return observation, or there might be doubt or uncertainty as to what is needed for a confident ID.

I appreciate it can be difficult when uploading audio direct from the app… you can’t edit the file to take out the silences or “frame” it to the most relevant section of the audio. I think the developers might be working on an audio editor for the app, or at least looking at how to utilise external audio editor apps, but I could also be wrong about that. Personally, I only upload audio in the browser, and only after having used an editor to trim and clean it up (in my case I use audacity).


@kiwifergus said it well. On iNat, the media evidence is primarily there to help others evaluate the observation and make an ID (although documenting behavior is also important, of course), so it shouldn’t be bigger than is necessary for that. We don’t support uploads of over 20mb, and we resize large photos to a maximum size of 2048 x 2048 pixels.

@jamesray would a few seconds of the calls be enough for others to evaluate when making an ID?

I just uploaded a 4 s audio extract. For me, I do enjoy observing their behaviours as a hobby, although obviously behaviours are useful and important for scientific purposes. However, it seems posting audio recordings for behavioural analysis may be better suited for xenocanto or eBird, although I do like the app upload functionality with iNat!

1 Like

When I’m out at reserves or places I expect to encounter birds etc I will often use an audio recorder on my phone and I start recording as soon as I get out of the car. When I hear a bird that I want to make an audio recording of, I hit stop on the recorder, and immediately hit start again. I then take a photo of the tree or bush the bird was in, even if I can’t see the bird (but of course I try and get the bird in photo too!). When I get home, I transfer the audio files to computer, open in editor, and the bird audio will be at the end of each. I frame it, clean it up with a levelling and noise filter, and then save it with the name being the date/time of the NEXT recording, which is effective observation date/time. I then upload it to the observation that has the photo (and should be same day/time too).

You can write up in the description, and link to external stored files if there are large audio files where the whole recording has interesting content, and same for video too. One of my interests is interactions, as well as lepidoptera raising on to link up obs of larvae and adult. Not technically what iNat is for, but as long as I am considerate of the primary mission of iNat, such off-label use can augment observations and “add value” without impeding on the primary mission.


Personally I find it more convenient to use Recforge II to record and edit all on my phone. I don’t bother with levelling and filters as it can adversely affect the audio quality, remove valuable environmental information, or just be more time-consuming. Again, cropping a lot can remove valuable information, although it’s true it does make it more convenient to listen to the actual vocalizations.


Rather than blocking all uploads if the file size is too large, the uploader should issue a warning that the audio file size limit is 20 MB, and the audio files are too large, then put that file (not the whole observation) in a staging area and continue to upload other files and observations. That will help to avoid issues being posted like this one.

1 Like

I think it would! Maybe write it up as a feature request, so that it gets considered by the development team :) If you do, you could add a link in the feature request back to this topic.

Can you please send some of the larger sound files to Our Android developer said he’d like to take a look.

1 Like

Please see the common myna folder here, along with many other recordings:!Yh8wxIJQ!R0bnR5URYuyoL6iCwTaNyg

Direct link:!10VB2C7b!3qipCu1cp3HSgjxC0dlnHA

1 Like