Node.js pipeline in the console for spectrograms creation and Birdnet detection for your Audiomoth sessions.
It will create a movie per audiofile with a spectrogram as image and utilizes birdnet AI to detect bird species.
The HTML page lists all movies and renders the birdnet detections.
Example result: https://audiomoth.ivo-bathke.name
You will need docker
,sox
and ffmpeg
on your machine.
On Linux install it like this:
sudo apt install sox ffmpeg
For docker installation see here: https://docs.docker.com/engine/install/.
Birdnet detection software will run inside a docker container.
Install:
npm install --global audiomoth-visualizer
Run:
audiomoth-visualizer <path-to/my-audiomoth-session> <destination-dir> --title="Session title" --locale=de-DE --timezone=Europe/Berlin --lat=50.8754444 --lng=6.8380488
Run:
npx audiomoth-visualizer <path-to/my-audiomoth-session> <destination-dir> --title="Session title" --locale=de-DE --timezone=Europe/Berlin --lat=50.8754444 --lng=6.8380488
Using npx you can run the script without installing it first.
Clone this repository:
git clone [email protected]:ivoba/audiomoth-visualizer.git
Install dependencies:
npm i
node index.mjs <path-to/my-audiomoth-session> <destination-dir> --title="Session title" --locale=de-DE --timezone=Europe/Berlin --lat=50.8754444 --lng=6.8380488
Then open the generated files/my-audiomoth-session/index.html
in your favorite Browser by doubleclick or via console:
chromium <path-to/my-audiomoth-session>/index.html
There are entrypoints to test birdnet AI or HTML generation separately:
HTML generation:
node_modules/.bin/zx html.mjs <path-to/my-audiomoth-session>
Birdnet detection:
node_modules/.bin/zx birdnet.mjs <path-to/my-audiomoth-session>
npm run release
Heavily inspired by:
https://github.com/nwolek/audiomoth-scripts
- responsive images
- responsive video
- with https://www.tensorflow.org/js/
- docker image in npx? will the image be pulled?
- docker user permissions on birdnet files
--user $(id -u):$(id -g)
user should be used in docker image kahst/BirdNET#23 - tests: https://dev.to/shadowtime2000/testing-command-line-tools-47bk
- https://engineering.britebill.com/2018/10/30/end-to-end-testing-cli-apps-jest.html
- print percentage of detection
- create a clean task to delete all files with no detection
- scroll to top link on bottom
- paging?
- fix Changelog generation
- input date -> output date options