Repository navigation
🧻 Mendil 0.2.0
A web panel, and a second, experimental voice engine that works without the TeamSpeak client.
Added
- Web panel on port 8080, laid out like a radio station's running order:
- The current track sits under an ON AIR lamp, with the time left.
- Every queued song shows when it will air, and the panel shows when the queue ends.
- From the panel you can pause, skip, stop, change the volume and the loop mode, request songs by link or search, remove, shuffle and clear the queue, and read the live log.
- It works on phones, in light and dark mode, and in English and Turkish.
- It is protected by a password (
PANEL_PASSWORD, or generated on first start).PANEL_TRUST_PROXYis for running it behind a reverse proxy. See README "Web panel".
- Native engine (experimental):
VOICE_BACKEND=nativeconnects to TeamSpeak without the official client, through teamspeak-js. It runs on x86_64 and ARM64 (ghcr.io/ulaski/mendil:native), uses about a quarter of the memory, plays music in stereo in any channel (no Opus Music codec needed), and logs why a server refused it (wrong password, ban, security level). It raises its identity's security level when a server asks for more. Classic stays the default. NATIVE_BITRATEsets the native engine's Opus bitrate.- CI tests the native engine end to end against TeamSpeak 3 and TeamSpeak 6 servers.
Changed
!searchonly lists playable tracks, not channels or playlists.- The controller is split into an engine-agnostic bot and two voice engines (
src/voice/). Classic behaves as before.
Update
docker compose pull && docker compose up -dNo setting has to change. To open the web panel, publish its port. See the README section "Web panel".
Try the native engine
Use the image ghcr.io/ulaski/mendil:native, which runs on x86_64 and ARM64. The README section "With the native engine" has a compose example. The native engine keeps its own identity, so give the bot its server group again after switching. To go back, use ghcr.io/ulaski/mendil:latest.