Shazam

Song recognition in the DankBar, powered by SongRec.

Requires

• songrec on $PATH (sudo pacman -S songrec on Arch)
• parecord and pactl (from libpulse, already there on a PipeWire setup)

How the sampling works

SongRec's own live capture does not work against a PipeWire monitor source: it samples for minutes without a single match on audio it recognizes instantly from a file. So the plugin does the sampling itself — parecord records a 13 second chunk from the chosen source, songrec recognize --json identifies the file, and that repeats until the listening budget runs out. A hit is reported the moment it happens.

Use it

• Left click the pill — opens the panel and (by default) starts listening.

• Right click the pill — starts or stops listening without opening anything.

• Panel — cover art, title, artist, album and year; one chip per service Shazam knows about for the track (Shazam, Spotify, YouTube Music, Deezer, Apple Music, plus a YouTube search); a copy button for "Artist — Title" and a second one for the full details (album, year, genre, ISRC, link); and a source switch, so output/mic can be changed without opening settings. Recent matches are listed below; click one to open it, right click to copy it.

• Custom action — set a command in the settings and a button appears in the panel. Nothing ships by default. The track is passed as arguments rather than spliced into the command: $1 artist, $2 title, $3 album, $4 Shazam URL, $5 "artist title". For example printf '%s\n' "\$5" >> ~/Music/wishlist.txt.

• Keybind — dms ipc call shazam identify (also stop, toggle, state, last), e.g. in Hyprland:

  bind = SUPER, S, exec, dms ipc call shazam identify
  

What it listens to

By default: system output. Each run resolves pactl get-default-sink and records that sink's .monitor source, so it follows your output device when you switch headphones or speakers — and it hears what the machine is playing, not the room.

Switch Listen to in the settings to Microphone for the room instead, or to Specific device plus an Audio device name from songrec recognize -l to pin one source.

Notes

• Recognition runs in a daemon surface, so a multi-monitor setup shares one songrec process and one result across every bar.
• A run is wrapped in timeout — SongRec would otherwise keep sampling the microphone forever when nothing matches.
• History is kept in the plugin state file, not in your settings.