conda-forge for BirdNET-Analyzer?

[cbalantic]

Hello!

I was previously using BirdNET-Lite to process audio files through BirdNET via R. I wrote up an install-BirdNET-for-Windows guide and documented some R functions that helped me process and wrangle the BirdNET-Lite outputs.

Now that BirdNET-Lite is deprecating, I am updating my workflows to accommodate BirdNET-Analyzer. With BirdNET-Lite, what worked to run via R was to set up a conda environment for BirdNET.

Now as I tinker with BirdNET-Analyzer instead, the error messages I'm getting lead me to believe that I will need to have my conda environment include the BirdNET-Analyzer-main module that I've locally installed. I'm not sure how to do this, since conda develop seems to be deprecated.

My question is, would it be possible to include your releases of BirdNET-Analyzer-main on conda-forge so that it could be easily installed to a conda environment?

(i.e., conda install - n birdnetenv -c conda forge birdnet-analyzer)

All of this may seem cumbersome to the more computer science-minded folks here, but I am trying to come up with well-documented ways to use BirdNET and promote access for more R-minded colleagues, collaborators, students, and interns who I work with. Apologies for the potentially uninformed question since Python is not my strong suit! It's possible there is a very easy solution if I were only using my own machine, but I'm hoping for a more portable solution so that I can easily document and explain to collaborators.

Thank you for what you do.

[cbalantic]

Update: in terms of an easy solution on my own machine (Windows), I got BirdNET-Analyzer to run from R in a conda environment by manually copying the "checkpoints" folder and eBird_taxonomy_codes_2021E.json file from BirdNET-Analyzer-main over into the conda env folder itself. Not elegant, but it's working... and it's way faster than BirdNET-Lite!

Still curious about the conda-forge question though. Thank you for considering.

[kahst]

I have to admit that I haven't worked with Conda so far, but releasing this repo as some kind of package seems to be a neat thing. I'll have to do some digging and can't promise anything right now, but I'll try. In the meantime: We are also working on an R interface for BirdNET which we will release with this repo (hopefully very soon). Ideally, this would mitigate some of the difficulties when setting up and running BirdNET for the non-CS community. I'll keep you posted.

[cbalantic]

Hi Stefan, sounds great. Looking forward to learning more.

[Josef-Haupt]

As mentioned in the README the current recommended way to install this repo is using the pypi package.
So it should be possible to use the pip in your environment to install the package locally.