Loudness & Dynamics

What it is

The finishing stage. Take a rendered song, optionally compress it, hit a loudness target under a true-peak ceiling, and get told what happened. It's the node every Plenio template ends in, and the one that decides whether your song sits at a sane level on Spotify or arrives as a quiet FLAC nobody turns up.

If you've ever bounced a mix and then guessed at a limiter setting, this replaces the guessing with numbers: BS.1770-4 integrated loudness, EBU loudness range, and a 4× oversampled true-peak meter, reported before and after.

How it works

The chain is fixed and documented: resample (if you chose a new rate) → compressor → make-up gain → true-peak limiter → measure, repeating the gain at most three passes to land within ±0.3 LU of the target. The resample-first ordering matters - converting after limiting creates new inter-sample peaks, so the ceiling would be a lie at the exported rate.

The compressor is stereo-linked with a soft knee and a side-chain high-pass (default 80 Hz) so the bass doesn't pump the whole mix; attack and release are smoothed and gain is computed at a 1 ms control rate. The limiter is 4× oversampled, has lookahead (3 ms by default) and limits true peak rather than sample peak. It's all ported from the author's earlier Music Production Toolkit and verified sample by sample against it - a difference below 2·10⁻⁶ of full scale, if you care about that kind of receipt.

target is a picker of real delivery specs, not vibes: streaming (-14 LUFS, -1 dBTP) as the default, Apple Music / podcasts (-16 LUFS, -1 dBTP), dynamic (-18 LUFS, -1 dBTP) for acoustic and classical, broadcast EBU R128 (-23 LUFS, -1 dBTP), and loud (-9 LUFS, -0.5 dBTP) for club and download masters. custom reveals lufs (-30 to -5) and ceiling_dbtp (-12 to -0.1).

compression is the style list - Balanced - gentle glue, Acoustic - gentle, Classical - preserve dynamics, Jazz - natural, Vocal pop - smooth, Pop - punchy, Rock - bus glue, Electronic - tight, Club - loud, Hip-hop - low-end punch, Ambient - spacious, Limiter only - transparent - plus off (limiter only) and custom with threshold, ratio and the rest. Each style bounds two things: make-up gain (18 dB default) and limiter reduction (6 dB for Balanced). Hit a bound and the node returns the best result within it and tells you which budget you ran out of rather than squashing your song to hit a number.

sample_rate (keep / 44100 / 48000) is the output-rate conversion, done before dynamics so the ceiling holds.

Outputs: audio at the chosen rate, and report - measured loudness and true peak before and after, plus each pass. That report is what you wire into Export Release so the release record carries your actual delivered numbers.

Install

# ComfyUI Manager → search "Plenio Music Production System" → Install → restart
# or:
cd ComfyUI/custom_nodes
git clone https://github.com/jplenio/Plenio-Music-Production-System

ComfyUI 0.37.0+. No Python packages installed by the pack - the DSP is the author's own, in NumPy, on the CPU - and no GPU is needed, so 4 · Enhance & Master runs on a laptop.

Common issues

• "Limiter reduction budget" warning. Reaching the target would need more limiting than the style allows. Pick a lower target or a style with more limiter room. The node deliberately doesn't flatten your song to satisfy a spec you didn't think about.
• "Samples above full scale were clipped." You boosted somewhere upstream without a limiter and exported FLAC/MP3. Keep this node (or the whole Master block) before the export.
• "Not measurable." Audio shorter than one 400 ms block, or silent, can't be loudness-measured; no make-up gain is applied and the limiter still holds the ceiling. That's the correct behaviour, not a bug.
• It's not louder than it should be. Good. Loudness is whole-song, with no per-section target, and there's no multiband compression, no stereo widening and no fades in this pack - if a song needs restoration, Plenio isn't the tool and says so plainly.