Quick Start Guide?

[evolsoulx]

Alrighty, this is like, exactly what i need. but i'm more a front end web dev guy. I'm literally stuck at the very first part of this, as in i need to compile it, but i'm just clueless here.

Tried using gpt to help a bit, but i think someone else could explain it quicker, and also, i think a lot of people could use a quick start guide?

Any way, for me specifically, i have access to a completely mine windows 11 machine, and a work limited mac. I think i can install whatever i need on the mac, and the windows machine is open. So what all do i need here?

[isZumpo]

Hi,

I agree that the documentation for getting started in terms of development could be improved. Do you have some specific steps i mind? Assuming that you want to make changes to the repository rather than just running the application? Because if you just want to run it, then I do recommend looking at the docker section of the readme.

If you would like to build the project for development purposes. Then the easiest metod to figure out how to build locally might be to take a look at the existing documentation as well as the dockerfile where all the dependencies etc can be seen.

What I personally have been doing is to make use of the dockerfile to build local changes as well. Do this in a terminal inside the root of the repository:

 docker build .

Take the docker hash/sha that gets spit out at the end and enter it into the docker run command:

docker run -ti \
  -p 8080:8080 \
  --device /dev/snd \
  <sha output from docker build>

(see readme for more options in the docker run command such as volumes)

You should now be able to visit localhost:8080 and see your local changes. Then whenever you modify something you will have to repeat the above steps again.

I hope that any of the two steps above help you gets started. Of course, feel free to ask if there is anything else too.

[evolsoulx]

TIL what docker is 🤣.

Ok i've got the sha, is this what we say when a program is built and signed? So i just did that for myself right?

So i'm stuck on this next docker run part. This is 100% me just not understanding docker, i swear once i can get it going i can do something worthwhile haha.

where is that docker run command actually... being run? windows cmd, mac terminal, is there some kind of console i'm not immediately seeing in docker?

doh. the new line for windows command is ^, getting further, system cannot find the path, obviously now i need to update those lol.

... and now i remember powershell is a thing.

powershell command for windows to list audio devices:
Get-WmiObject Win32_SoundDevice | Select-Object Name, ProductName, DeviceID, Manufacturer

[evolsoulx]

hah yeah, at this point, i can't even get it just up and running. 'something' here doesn't make sense, but i don't know what's blocking me

[isZumpo]

So it appears that there are many things happening at once here then :)

where is that docker run command actually... being run? windows cmd, mac terminal, is there some kind of console i'm not immediately seeing in docker?

Docker is a command line utility like any other. Meaning that you will be executing it in your terminal of choice.

First of all it might be a good idea to verify that you can run docker at all on your windows setup. Maybe follow some sort of beginners guide to get started. Windows is a bit tricky and it might suggest you to install WSL etc, leaving you wishing for that you had just installed Linux rather than Windows from the start... Probably easier to install docker on your mac if you just want to get started quickly.

Once you have verified that you can run some sort of docker image on your computer and feel ready to get started with birdnet-go. This should be the most simple docker command for that:

docker run -ti -p 8080:8080 --privileged ghcr.io/tphakala/birdnet-go:latest

(Adding the privileged flag above since I am not sure how mounting of audio devices works in windows...)

[tphakala]

I finally managed to get Github Workflow working for automatic build for releases, took me several days to get it going .. Anyhow there is now a fresh binary release for Windows at https://github.com/tphakala/birdnet-go/releases/tag/v0.5.0

Zip contains two files, birdnet-go.exe which is a native Windows binary, (no need to use Docker or WSL), and tensorflow lite c api library libtensorflowlite_c.so, which must be included in system PATH or in same directory birdnet-go.exe is being run at.

Run .\birdnet-go.exe once to get config.yaml file created and you can configure it with notepad etc. To enable log which you can use for OBS chatlog set

realtime:
  log:
    enabled: true
    path: path\to\chatlog.txt (this is file which you refer to in OBS)

To make spectrograms work you need to install SoX which is available for Windows at https://sourceforge.net/directory/?q=sox-14.4.2-win64.exe

Run .\birdnet-go.exe realtime and it should start listening on default audio capture device on Windows, if it attaches on wrong audio device you need to change default recording device in Windows sound settings. Web UI is available at your Windows system IP on port 8080.

[evolsoulx]

!!! Up and running! I was worried with the whole audio config thing, but you have a simple and working way for me to hook my rstp stream in with no issues!

I can't seem to get sox to work, got it installed and path set up (i can do sox --version in powershell, but beyond that, dunno), but i don't need that, the log file is what i need, and it's absolutely perfect <3.

This whole week i've been trying to set up a new box to do some things on. It's got a wonderful bloatware Acer install of Windows 8.1, which i had never used. I'm in the process, finally, of reinstalling everything and finally have it working, but once i can find a USB drive i'm going to throw linux on it and try to get this all running there.

Incredibly stupid question; Ubuntu 23 is what i want, yeah? I've never actually set up a linux box, always sounded so complicated, but looks like it's stupid easy now. That is if i can find this damn usb stick. Going to end up just buying one when i know i have like 30 lying around somewhere 🤣

Thank you so much for your help, like i said, once i figure out all this set up crap, i can probably give you some cool frontend stuff.

[evolsoulx]

https://www.twitch.tv/squirrelville got a super first-go display; just a python window parsing the last 5 of the text file, but oh my god, you have no idea the smile on my face right now :D

and another dumb question: the sensitivity; is lower detecting more or less sound?

[tphakala]

Sensitivity value is a setting of BirdNET model and I can't say that I fully understand it, in technical terms it is "sigmoid sensitivity" and it is part of neural networks algorithms used in machine learning. If you are seeing too few detections you may increase it and see if helps, but increasing its value will most likely also increase number of false detections.

About what Linux to use, yes I guess Ubuntu Linux 23.10 is a solid choice,

[evolsoulx]

got ubuntu installed! gonna play with that this week and get it set up as a 'dev' box, initially just for this, but i need to figure all this stuff out, it's holding me back.

is birdnet using the same 'stuff' as ebird in terms of identifications (new to bird call/watching world, getting terms down lol)? I'm missing a lot of cardinals that Merlin is picking up from the same audio feed, but it could easily just be a sensitivity/settings thing.