-
Upload the mix
A stereo mix, bounced without a limiter on the master bus if you have the choice.
Nothing is re-encoded on the way in, so a 24-bit 96 kHz WAV is analysed as a
24-bit 96 kHz WAV.
-
Measure what came in
Before anything is changed, the mix is measured: integrated loudness, true peak,
loudness range and spectral balance. That measurement is kept and shown beside the
finished file, labelled as the input, so you can see what moved.
A master and a mix are never shown as one unlabelled number. A −20 LUFS mix
mastered to −14 that still read “−20.0 LUFS” would tell you
nothing happened.
-
Set the tone from your mix
The tone target is your own mix with one small curve added: the median change a
working mastering engineer made across 17 masters from one soundtrack album. The
chain corrects toward that target as it limits the mix, and leaves the arrangement alone.
You choose nothing here and there is nothing to get wrong. The same curve goes on
every mix, so the result depends on your audio rather than on how you would have
classified it.
-
Render to the delivery target
The mix is mastered toward the target you chose. Both WAVs are held to the
true-peak ceiling. Three files are written: the 24-bit master at your source sample
rate, a 16-bit 44.1 kHz version, and an MP3 at 320 kbps.
The chain evaluates candidate gains, measures what the limiter actually did to each
one, and keeps the best result it has seen — never the last one it tried.
If the mix could not reach the target without passing the ceiling, the file is
still delivered — at the loudest level that stays compliant — with the
shortfall stated in LU. The ceiling is never traded away, because the ceiling is
the reason the target was out of reach.
-
Measure the file that will be delivered
Every rendered file is decoded and checked again, for length and for invalid
samples. Both WAVs are held to the true-peak ceiling, and the 24-bit WAV’s
loudness is checked against the target, or against what mastering reached when the
target was out of reach. That catches a botched resample or a corrupted render
before it is offered. The MP3 is decoded and measured too: at the default
−1.0 dBTP ceiling, one that would peak above 0.0 dBTP is encoded
again slightly quieter, so it does not clip. At −0.1 dBTP it is delivered
as encoded, and its peak is reported rather than held.
Choose Dynamic (−14 LUFS), Balanced (−11 LUFS), Full (−9 LUFS), or Extra loud (−8 LUFS). Each is held at −1.0 dBTP by default, the ceiling Spotify, Apple and the AES recommend for anything that gets re-encoded. Switch to −0.1 dBTP for a WAV nobody re-encodes.
Technical reference: 18 supported delivery presets, including saved legacy targets
New uploads offer the four loudness choices above at −1.0 dBTP: Full (the default row) and streaming, loud-11 and loud-8 below. Each can be switched to −0.1 dBTP: streaming-tp01, loud-11-tp01, loud-9-tp01 and loud-8-tp01. The other presets remain supported for saved jobs and retries, including their original output formats.
Scrolls sideways
