Changelog

v0.5.0

(ui) Web UI for data visualization
(detection) Privacy filter to discard audio clips with human vocals
(detection) Dog bark tracking for common false positive filtering
(db) Save all BirdNET prediction results into table Results
(audio) Check user group membership on audio device open failure and print instructions for a fix
(docker) Added support for multiplatform build
(conf) New function to detect if running in container
(audio) rtsp audio stream capture support by @twt--

(docker) Install ca-certificates package in container image
(capture) Set capture start to 5 seconds before detection instead of 4 seconds
(capture) Increase audio capture length from 9 to 12 seconds
(rtsp) Wait before restarting FFmpeg and update parent process on exit to prevent zombies

(database) Switched sqlite journalling to MEMORY mode and added database optimize on closing
(workflow) Update GitHub Actions workflow to build and push Docker image to GHCR
(workflow) Support multiplatform build with github actions
(docker) Add ffmpeg to container image
(ui) Add Greek (el) translations by @hoover67
(ui) Improve spectrogram generation to enable lazy loading of images

(ui) Improved spectrogram generation
(database) Move save to common interface and change it to use transaction
(audio) Stopped audio device is now simply started again instead of full audio context restart
(audio) Set audio backend to fixed types based on OS, default to ALSA on Linux
(labels) Update makefile to zip labels.zip during build and have label files available in internal/birdnet/labels to make it easier to contribute language updates
(audio) Improve way start time of audio capture is calculated