The plugin and the standalone app

LoudSense also runs natively: as a VST3 and AU plugin for the master bus and as a standalone app for files. Both use the same measurement core as this browser tool, checked against the same reference signals, so a file reads the same everywhere. Audio passes through the plugin untouched.

Meter

Live LUFS-M, LUFS-S, LUFS-I, LRA, true peak and clips since the last reset. Pick a target group and a target row: the bars show its window, the true-peak bar its ceiling, and the value turns amber or red when you leave them. Gain to target is the number to type into the host's output trim; Copy puts it on the clipboard. Peaks since reset logs every clip run and every true-peak overshoot with its time, so after a long session you know where the three overshoots were.

Every verdict says what to do. Peaks too hot tells you whether turning the whole mix down by the overshoot keeps the loudness inside the window (then the gain box offers that number) or whether you need a true-peak limiter on the master at the ceiling instead.

With Reset on locate (on by default, on the Timeline tab) the live run restarts whenever playback starts from a new position, so LUFS-I is the pass you are hearing, not the last hour.

Timeline

The song, not the pass. Every block the host plays through LoudSense lands on the song position the host reports, in 100 ms slots. Play any part in any order: the newest pass over a part replaces the older one, and the numbers and verdict rows are stitched from everything measured so far, with the same gating as a single pass. One contiguous pass over a song gives exactly the numbers a capture would.

The bar shows what is covered and how fresh it is: bright for the last two minutes, dimmer for older. The list next to it shows each measured stretch with its age. Double-click a stretch to clear it; Clear starts over.

The plugin cannot see edits. After you change a fader or move a region, play that part again and the new pass replaces the old measurement.

Analyse a part of the song

A plugin never gets the timeline itself from the host, only the audio the host plays through it. So the part has to play through LoudSense once. Three ways, pick what fits how you work:

1. Play it in realtime. Select or cycle the part and press play. It lands on the Timeline; with Follow transport it also becomes a capture.
2. Offline bounce of the cycle range. Bounce the selection with LoudSense on the master. The host renders far faster than realtime and still reports the position, so the part is measured in seconds. Throw the file away afterwards.
3. Export the selection into the watch folder. Same bounce, but keep the file: the standalone app measures it as it lands and raises a notification when it misses the target.

Capture

Press Capture (or R), play the pass, press Stop. The Panel tab then shows what the browser tool shows for a file: the waveform with clip and peak marks, the loudness curve, the stats, the speech/music line and the verdict rows. The last five passes are kept and listed in Compare, so "before the limiter change" sits next to "after".

Follow transport (Timeline tab) does the pressing for you: play starts a capture, stop finishes it. A capture that ran faster than realtime, an offline bounce, is named Bounce.

Markers. Return (or the Marker button) drops a marker while capturing; shift-click the panel adds one afterwards; Import markers reads a marker export from the DAW (seconds or m:ss, tab or comma, name after it). The Sections table then reports LUFS-I, short-term max, LRA, peak and clips per section — verse, chorus, ad read.

Files (standalone)

Drop files or folders: a row per file with the verdict for the chosen target, the album loudness underneath, and Report CSV / Report HTML with a row per file and a column per platform. The HTML report prints to PDF from the browser.

Watch folder points at the DAW's bounce folder. Every new export is measured when it has finished writing; a bounce that misses the target raises a notification. Play / Hear as plays a file at the level a platform normalises it to.

History and targets

Every capture and file is remembered on this machine, numbers only. Pick a row in History to see the trend of its series. Targets holds the built-in table plus your own: add a station spec or a client rule, export it as a .lstarget file and send it to the mix engineer, who drops it on their Targets tab.

Dialog-gated loudness

Next to the full-mix value the plugin shows the integrated loudness over the parts the speech detector marks as speech, the measure Netflix (−27 LKFS) and Apple TV+ (−24) use for delivery. LoudSense's detector is its own and the row says "estimate"; the platforms use theirs.

The numbers

What Peak, True peak, Clips, LUFS-M, LUFS-S, LUFS-I and LRA mean is on Reading the numbers. The same explanations are in the plugin's ? panel, together with the keys: R to capture, Return for a marker.

Install

macOS: the VST3 goes to ~/Library/Audio/Plug-Ins/VST3, the AU to ~/Library/Audio/Plug-Ins/Components, the app to Applications. The build is not notarised by Apple yet, so macOS blocks it once on first launch; the install notes in the download carry the one-line Terminal command that clears this. Windows: VST3 and standalone, unsigned; SmartScreen asks once.