findingfocus.music

Waveform music player built with SvelteKit 5 (static adapter) and wavesurfer.js. Audio is streamed straight from Cloudflare R2 (bucket findingfocus-music).

Local dev

npm install
npm run dev

Checks & build

npm run check    # svelte-check
npm run build    # builds static site into /build (adapter-static)
npm run preview  # serve the production build locally

Local testing with Docker

docker compose up --build   # -> http://localhost:3000 (container: ffmusic)
docker compose down

Recording in SuperCollider

// REC TOGGLE — run once to start, again to stop
(
var numChans = 2;
var dir, path;
s.waitForBoot {
  if (s.isRecording) {
    s.stopRecording;
    "STOP -> % (exact take)".format(~recPath).postln;
    ~recPath = nil;
  } {
    dir = PathName(thisProcess.platform.userHomeDir) +/+ "recordings";
    path = (dir +/+ (Date.getDate.stamp ++ ".wav")).fullPath;
    s.recHeaderFormat = "wav";
    s.recSampleFormat = "int16";
    s.recChannels = numChans;
    s.record(path: path, numChannels: numChans);
    ~recPath = path;
    "REC ON -> %".format(path).postln;
  };
};
)

Publishing a track

Drop a take in $RECORDINGS_DIR (default ~/recordings), then run the publish script. It picks the newest .wav, renders peak data, encodes the MP3, uploads it to R2, and writes the resulting track straight into the live tracks.json — no code edits or deploy needed.

bin/publish-nightly.sh                # publish the newest .wav in ~/recordings
bin/publish-nightly.sh --wav ~/take.wav

Options:

Flag	Description
--wav FILE	WAV to publish (default: newest .wav in $RECORDINGS_DIR)
--code FILE	Tidal code to attach (default: <date>.tidal in $FF_TIDAL_DIR, else pulled from findingfocus.io/findingfocus/tidal)
--title NAME	Override the track title (with --replace: renames the existing track)
--fade N	Loop crossfade width in seconds (default 0 = raw take; >0 wraps)
--edge-fade N	Fade in/out at file edges (default 0.05; use 0 to disable)
--replace SLUG	Re-encode and overwrite an existing track in place, e.g. 2026-08-29_1
--dry-run	Build everything and print what would upload; upload nothing

The bucket must allow cross-origin GET requests (CORS GET, origin *) so the browser can fetch and decode audio for the waveform.

Deploy (Kamal)

Config in config/deploy.yml.

kamal setup    # first time (install/configure the web server)
kamal deploy