libsonare

English

日本語

English

日本語

Appearance

Mixing Demo Project JSON ​

The Mixing Studio saves its whole mix state as libsonare-mix-project.json: tracks with their clips, strips, sends, outputs, and automation, the two aux buses, the two subgroup buses, the master, VCA gains, and the loop range. The file excludes decoded audio. Use Save project JSON to write it and Open project JSON to read it back; the tracks come back waiting for audio, and each file you add attaches to the track with the same name, otherwise to the next one in order. For the app architecture and source files, see Build a Browser Mixer.

This is not the mixer scene format

This is a different exchange format from the Mixing Scene JSON reference. A scene describes a whole mixer graph — strips, buses, VCA groups, and connections — and round-trips through Mixer.toSceneJson() / fromSceneJson(). The demo project file below is the state of a fixed-layout console, read only by the browser demo; the engine's fromSceneJson(...) does not accept it, and the demo rejects a scene file.

The demo below is the engine's own lane mixer rather than the /mixing page, but it plays what a project's tracks describe: three tracks, each with a fader and a mute. Each band is one entry in tracks[] — the fader is that entry's faderDb and the mute is its muted, and the demo hands both to the engine through the scene format's strip (strips[].faderDb) and setSoloMute. What the demo has no counterpart for is clip.startSec: its clips all start at zero. Placing a track later on the timeline is what the project file adds on top of the strip settings the engine already understands.

ENGINE · LANE MIXERIDLE
The engine lane mixer — faders and mutes inside the playback engine

Three MIDI clips loop through the realtime engine: each track occupies a lane with its own channel strip, and the lanes sum into a master strip carrying the engine’s true-peak limiter. The faders call the strip setters and the mutes call setSoloMute — every band below is the engine’s separate solo previews through the master chain, re-rendered through renderOffline as you move the controls.

Lead fader
0 dB
Bass fader
3 dB
Drums fader
-8 dB
Mute lead
Mute bass
Mute drums

Fixed layout ​

Every track has the same strip: six EQ bands, then six insert slots in order — trim, gate, comp, deess, duck, width — and two sends, to the reverb and delay buses. The master has two slots, glue and limiter. A bypassed slot stays in the file, so slot positions never change. Positions in the file are seconds; the engine runs at 60 bpm, so one second is one quarter note.

Shape (version 2) ​

The example is abbreviated: a real file lists all six EQ bands and every slot parameter.

json
{
  "version": 2,
  "tracks": [
    {
      "id": 1,
      "name": "Lead Vocal",
      "sourceDurationSec": 42.5,
      "clip": { "startSec": 1.5, "trimStartSec": 0, "lengthSec": null, "fadeInSec": 0.02, "fadeOutSec": 0.5 },
      "faderDb": -3,
      "pan": 0,
      "panLaw": 0,
      "panMode": 0,
      "dualPanLeft": -1,
      "dualPanRight": 1,
      "channelDelayMs": 0,
      "polarityLeft": false,
      "polarityRight": false,
      "muted": false,
      "soloed": false,
      "soloSafe": false,
      "vcaGroup": null,
      "outputBus": null,
      "eq": [
        { "type": "HighPass", "frequencyHz": 30, "gainDb": 0, "q": 0.707, "enabled": false }
      ],
      "slots": {
        "trim": { "bypassed": false, "params": { "levelDb": 0 } },
        "comp": { "bypassed": true, "params": { "thresholdDb": -18, "ratio": 2 } },
        "width": { "bypassed": false, "params": { "width": 1 } }
      },
      "duckKeyTrackId": null,
      "sends": {
        "reverb": { "levelDb": -18, "timing": "postFader" },
        "delay": { "levelDb": -60, "timing": "postFader" }
      },
      "automation": {
        "fader": [{ "id": "p1", "timeSec": 8, "value": -6, "curve": "linear" }],
        "pan": [],
        "width": []
      }
    }
  ],
  "buses": {
    "reverb": { "gainDb": 0, "params": { "decay": 0.566, "damping": 0.5, "preDelayMs": 20, "dryWet": 1 } },
    "delay": { "gainDb": 0, "params": { "delayTimeLMs": 250, "delayTimeRMs": 250, "feedback": 0.25, "pingPong": 0, "dryWet": 1 } }
  },
  "groups": { "group1": { "gainDb": 0 }, "group2": { "gainDb": 0 } },
  "master": {
    "faderDb": 0,
    "slots": {
      "glue": { "bypassed": true, "params": {} },
      "limiter": { "bypassed": false, "params": { "ceilingDb": -1, "releaseMs": 50 } }
    }
  },
  "vcaGains": { "A": 0, "B": 0, "C": 0 },
  "loop": { "enabled": false, "startSec": 0, "endSec": 0 }
}
  • id is a positive integer, unique within the file. A send at or below -60 dB is not routed. lengthSec: null plays the rest of the source.
  • eq has exactly six bands. type is one of the engine's EQ band types (Peak, LowShelf, HighShelf, LowPass, HighPass, BandPass, Notch, TiltShelf, FlatTilt, AllPass).
  • curve is linear, exponential, hold, or s-curve, and shapes the segment up to the next point. Each lane is fader (dB), pan, or width.
  • outputBus is "group1", "group2", or null for the master; any other value is rejected. groups holds each subgroup bus's gainDb (default 0). A subgroup bus is a summing fader with no inserts. Both are optional, so a v2 file saved without them still opens, with every track on the master.
  • vcaGroup is "A", "B", "C", or null. duckKeyTrackId names the track whose signal keys the duck slot, or null to key it from the track's own input.
  • A file holds at most eight tracks.

Omitted fields take the demo's defaults. A present field of the wrong type, a slot parameter the slot does not have, a duplicate track id, a duckKeyTrackId that names no track or that keys a track from itself through a chain of keys, or an eq array that is not six bands rejects the whole file before anything changes. Numeric values must be finite numbers; a string such as "-6" is rejected.

Opening a v2 file replaces the whole project, undoably.

Reading a version 1 file ​

A v1 file is a flat per-track settings file with "version": 1 (or no version). The demo converts it onto the tracks currently loaded, so load the matching audio first. A notice says the file was converted.

json
{
  "version": 1,
  "masterFaderDb": 0,
  "tracks": [
    { "id": "track-id", "name": "Lead Vocal", "offsetSeconds": 1.5, "inputTrimDb": 0, "faderDb": -3, "pan": 0, "width": 1, "muted": false, "soloed": false }
  ]
}

A v1 file must carry at least one of masterFaderDb, reverb, vcaGains, or tracks; tracks entries need an id or a name, and every supplied control must have its expected type. An engine scene (strips, buses, or connections) is rejected. Anything that is neither a valid v1 nor a valid v2 file is refused.

The conversion follows these rules:

  • Tracks match by id or name. A v1 track that matches no loaded track is dropped, and the notice lists its name. Omitted top-level settings keep their current values, and a supplied track takes defaults for what it omits.
  • offsetSeconds becomes clip.startSec, kept inside the timeline. inputTrimDb becomes the trim level and width the width slot.
  • Tone becomes strip EQ. A tilt of t dB is a low shelf at −t/2 and a high shelf at +t/2, both at 1 kHz; air is a high shelf at 12 kHz in the second Peak slot. A track with eqEnabled: false gets the default, disabled EQ.
  • reverbSendDb becomes the post-fader send to the reverb bus; a v1 reverb.enabled: false turns those sends off. reverb.decaySec and reverb.preDelayMs set the plate's decay and pre-delay.
  • A v1 automation list, which held every parameter in one array keyed by param, splits into the fader, pan, and width lanes, sorted by time. An unknown curve becomes linear.
  • Values are clamped to the v1 ranges: faderDb to −60…+12 dB, masterFaderDb and each vcaGains entry to −24…+12 dB, inputTrimDb to ±24 dB, pan to ±1, width to 0…2, channelDelayMs to 0…50, and a send to −60…0 dB.