What are you doing with Chirpity?

[Mattk70]

I'm curious to know how people use the results of the audio they analyse. This will help me understand better what the best ways might be to develop the "after analysis" features of the software.

Here's a quick poll. Also - if you have a story you'd like to share - use the show and tell to let me know about it!

[Vollpflock]

Initially, I went for the "Share online" option, but it turns out the csv file is all I need and I don’t know that else I want from it. I’m currently reporting observations by using a script, which is going smoothly (especially since there’s a column for latitude and longitude).

A potential improvement could be a faster way to save detections—perhaps bypassing the record creation step with a keyboard shortcut. It would also be useful to have two categories for the observations: "confirmed" for the definite ones and "unconfirmed" for those that are uncertain and need further verification.

However, what I’m currently interested is in building my own sound library.

I haven't found another convenient method for creating a library that includes both sound files and their spectrograms and think Chirpity can help a lot in these regards. So, I'm leaning more towards the archive feature now.

I'm not exactly sure what I'm looking for in it, though. I need to test first how the .wav files hold up after being compressed. Do you normalize them before compression, and which compression format are you using?

Edit:

Just as a thought:
Maybe separating “Share online” and “Just keep them for my own use” more? Here's what I'm thinking: I go through the recordings and make new records for sounds I don't want to keep in the sound library, just for the sake of reporting them. In the end, my archive gets cluttered with a bunch of calls that I don't really want to keep but only created records for to report my observations.

I'd like to save the good sound clips for my library and the uncertain ones for later verification. Maybe we could have a shortcut for "reporting"? Like, adding an extra column in the CSV file labeled "Report" with "Yes" for the ones I'm reporting.

[Mattk70]

XC does have an API for download. You can make a request for the spectrogram, audio, and metadata fields. It shouldn't be so hard to make use of this. So, instead of the "Compare species on XC" link opening a browser window, it could open a new Chirpity window. The window could have a carousel of the XC species' media files loaded with the Chirpity spectrogram view. You could organise the files by NFC, Flight Call, Call & Song - credit the contributor / include field notes beneath the spec. Hmm...

[Vollpflock]

Interesting idea, need to think about it

[Vollpflock]

Wow, you are already into it - awesome

[Mattk70]

It is looking (and working) pretty well. Sneak preview here. Click the new image for a bigger view.

I'm working on eBird record format export, but I've run into issues with taxonomies. You'd think it'd all be joined up at Cornell, but apparently not...

[Vollpflock]

You never disappoint

[Vollpflock]

Jouuu and the comparison is even in the same color - awesome! Xeno-Canto was always a pain to navigate for me.

Btw, have you already gotten a response from Xeno-Canto considering the Upload-API?

If I was building a sound library, I'd post the examples to a site like Xeno-Canto. I've been burned by HDD failure too many times in the past. The problem with these sites is they don't have an upload API, so you have to upload files using a form on their site. I think, one at a time. In the last couple of days I've been talking to one of the founders at XC about an upload API - so we'll see what comes of that. Ideally, you'd be able to select detections to upload to XC (or eBird/BirdTrack/Trektellen) directly from Chirpity, which could format the audio optimally for their site and pre-fill most of their required data fields. Fingers crossed!

[Mattk70]

Nothing from the guys at XC so far. I'll nudge them in a few days.

[Vollpflock]

Thought I'd check in again. Have you heard anything from XC in the meantime? Edit: Just saw it.

[David-Birdbrain]

I'd like to share my use case, not least to confirm I'm using the appropriate software :)
I run a re-wlding project in France. My goal is to leave my phone outside ~24hrs to record bird song. Do this 4-5 locations. Then import these recordings and get a record of which birds are present. I will store these in iNaturalist
Then repeat this each quarter.
My goal is to compare and contrast the project with land outside the project and to track changes in the project bird population over time
Does Chiperty seem appropriate for this?
@Mattk70

[Mattk70]

Hi David, yes, I think Chirpity will be useful for this purpose. You will get candidate species IDs from the automated analysis, and need to verify these IDs. I'm not sure where in France your project is, but you may find some species you are interested in are missing from Chirpity's native model, since it is focused on species recorded in the UK. However, you can do your analysis using the BirdNET model too, if that is an issue. You can save your detections for each location and review changes over time, but it sounds like you'll be using iNaturalist for this aspect of your project.

[David-Birdbrain]

Thats great - Agree BirdNet was the model I was planning to use.
My main concern was whether the app could support multi- hour recordings

[calyptura]

First of all, I want to thank you for developing Chirpity. It has been a game changer in helping me monitor bird communities in Brazil using autonomous recorders. I’d like to share how I’m using it, as it might make the tool even more valuable for me and other researchers.

I’ve developed a protocol where I listen to 48 minutes out of a total of 72 hours of monthly recordings from each autonomous recorder deployed in the field. I manually annotate all the species I detect in the 48 minutes I listen, which usually covers the most common birds in the area. Afterward, I run all 72 hours of recordings through Chirpity to expand the species list by identifying rarer and less common species.

Next, I run individual pattern-matching models, searching for the calls of all species — both those I identified manually and those Chirpity detected — across the entire dataset (72 hours). Chirpity has also been incredibly useful for facilitating the pattern-matching process by helping me quickly locate the best-quality recordings for each species within the dataset and use them as models for pattern matching.

I also tried using Raven Pro and BirdNET-Analyzer GUI for the same tasks, but Chirpity has proven to be much more useful for the protocol I developed, as well as being far more user-friendly.

Once again, thank you!

Luciano

[Mattk70]

That's awesome, thanks for sharing!