Online mastering

Your master is safe in the WAV. Nobody plays the WAV.

Drop a track in and get a finished master back in minutes: a 24-bit WAV, a render per platform, and a report that shows what changed.

Every platform re-encodes your file, and the encoder moves the peaks. We encode to the real codec, measure what comes back, and lower the ceiling until what the listener receives is correct.

Master a trackHear it firstThree free every month, WAV included. No account, no card.

What lands in your downloads

One upload, four things back. The free tier includes all of them.

The 24-bit WAV

The master itself, at full resolution, yours to use anywhere. Not watermarked, not a preview, not behind the paid tier.

A render per platform

One file each for Spotify, Apple, YouTube and the rest, at the loudness and the peak that platform asks for.

The report

Loudness, true peak, dynamics and stereo width before and after, plus every repair that ran and why it ran.

A verdict per platform

Each render encoded to the real codec, decoded, and measured again. Green means it survives the encoder, not that it looked fine in the WAV.

Three steps, no install

  1. Drop in a track

    WAV, MP3, FLAC or M4A, up to 200 MB. A whole release works too: add the tracks together and they get one shared target.

  2. Pick a target, or none

    A style profile, your own reference track, or no EQ at all when the mix already sounds the way you want it.

  3. Read the numbers, take the files

    Compare against your source at matched loudness first. If it is not right, change one setting and run it again.

Master a trackNothing to install. It runs on our machines, on CPU.

Four things, four measurements

Not features. Each is the figure it produced.

Encoder lift

+0.86dB

Codec-safe mastering

Added by YouTube's encoder alone.

Tonal spread, 3 tracks

1.50→1.09dB

Album mode

Separate generations, one shared target.

Spectrum wall found

20.1kHz

A lossy source stays honest

Graded, never boosted above.

Decorrelated bass in mono

8.8→0.0dB

The low end, handled separately

Phones are mono. Under 120 Hz goes centre. Not from this repository

Interactive

Every level, measured through every codec

Nothing is computed in your browser. Each figure is a real encode, decoded and measured.

Master level, LUFS

headroomnone left

Came out
-16.00 LUFSon target
Worst after codec
-5.81 dBTPok TikTok
Ours
-5.81 dBTPok ceiling -6.55
Came out
-14.00 LUFSon target
Worst after codec
-3.69 dBTPok YouTube
Ours
-3.69 dBTPok ceiling -4.55
Came out
-12.00 LUFSon target
Worst after codec
-1.83 dBTPok YouTube
Ours
-1.83 dBTPok ceiling -2.55
Came out
-10.56 LUFS0.56 dB short
Worst after codec
-0.27 dBTPtight YouTube
Ours
-1.55 dBTPok ceiling -1.98
Came out
-10.07 LUFS2.07 dB short
Worst after codec
-0.26 dBTPtight YouTube
Ours
-1.51 dBTPok ceiling -1.98
Came out
-9.31 LUFS3.31 dB short
Worst after codec
-0.48 dBTPtight YouTube
Ours
-1.22 dBTPok ceiling -1.76
  • ok inside the limit
  • tight under full scale
  • clips over 0 dBTP

Every platform at every level, with codec, lift and playback

Before and after

Both players are level-matched

Louder reads as better, so an unmatched A/B only proves which file was pushed harder. Both play at -13.87 LUFS.

Source, unmodified

-13.87 LUFS
-4.41 dBTP

Master, level-matched

-13.87 LUFS
-1.44 dBTP
  • Source
  • Target profile
  • Result
Distance to target: 5.67 dB source, 2.87 dB result.

The whole run: every repair step, every platform verdict

Album mode: three tracks, one target

Tonal spread, per band
1.50→1.09dB
Worst single band
4.79→2.81dBat 400 Hz
Loudness spread
1.02→0.00LUnormalisation alone does this
Stereo width spread
0.044→0.000
  • Before, as generated
  • After album mode
Spread between the 3 tracks per band. Lower means they agree.

Per track, and what this run does not prove

8 profiles, one source, nothing else changed

All level-matched to -14 LUFS, so the profile is the only variable left.

The source, no profile

-13.08 LUFS
-4.38 dBTP
wall 20.2 kHz

Hip-hop and trap

5.35 → 2.05dB

EDM and dance

5.11 → 1.85dB

Lo-fi and chill

3.83 → 1.71dB

Acoustic and singer-songwriter

5.38 → 2.25dB

Rock and band

5.26 → 2.06dB

Pink reference

4.83 → 1.95dB

No EQ, loudness only

0.37 → 0.03dB

Every curve: ±9 dB, 40 Hz to 16 kHz. What each profile does, per band

Public reference

Delivery targets, and how much to trust each one

1 of 6 publishes its own loudness figure. Checked 2026-08-07.

The full reference, one anchor per platform.

Before anything is processed

Ask what is wrong with the mix first

A build-up in the low mids can be mastered, and then you have a mastered build-up in the low mids. The only place to fix that is the mix.

Is the source intact
A wall high in the spectrum is what a good encoder does and it costs you nothing, so it is not reported as a problem. A wall low enough to hear as dullness is, because no mastering brings that back — here or anywhere.
Is there room left to work with
Sitting against the ceiling, or clipped on the way out, means everything starts from what survived the export. Dense is not the same as limited, and the report does not treat them as one.
What would the master have to fight
The same findings the master report gives, before instead of after: harshness, an abrupt stop, a low end that collapses when a phone folds it to mono.
Check a mix

No file comes back, so it costs nothing and does not count against your free masters.

The WAV download is free, because paywalling it is the problem

Paid tiers buy volume, never better audio.

Free

Free

3 masters a month. The WAV download is included.

  • 24-bit WAV master
  • The full report
  • The codec-safe pass
  • Platform verdicts
  • Mix checks that do not count
Master a track

Single track

$2per track

No account. Pay for one track, download it, leave.

  • Everything in Free, one track
  • No sign-up, nothing to cancel
  • All platform renders
Master one track

Unlimited

$12per month

No limit, plus the parts that only matter once you release regularly.

  • Unlimited masters
  • Album mode across a release
  • The distributor package
  • Your own reference as target
Go unlimited

Opening prices. EU customers pay VAT on top. Terms.

The one listening test

Eight rounds, blind, at matched loudness

Louder reads as better, so the only test worth doing is one where nothing is louder.

Mastered with a profile7/8
Mastered, no tonal matching1/8
The original, unmastered0/8

The unmastered file was chosen 0 times out of 8. All three versions played at -16.0 LUFS with the peak under -1.0 dBTP, so loudness was taken out of the comparison before it started. The names and the order of the buttons changed every round. Guessing would have landed each version around 2.7 times; the odds of one version taking 7 of 8 by chance are 0.8%, corrected for there being 3 of them.

And what it does not prove: one track, one listener, and that listener built the engine. He did not know which button was which, but he knew what kind of difference to listen for, so this is blind and not double-blind. It is enough to answer whether the tonal matching does anything. It is not enough to claim this sounds better than anyone else, and that claim is not made anywhere on this site.

The honest part

What this cannot do

Worth reading before you spend anything, because these are the cases where the answer is no.

We cannot bring back what the encoder threw away

Above the wall in your spectrum there is only noise to amplify. We print where the wall is and leave it alone.

We do not make stereo out of mono

Widening means phase tricks, and phase tricks collapse on a phone speaker. We leave it mono and say so.

We master, we do not mix

A buried vocal cannot be lifted without lifting everything at that frequency. Fix the balance where it lives.

We do not distribute

You get the WAV and a render per platform. Getting it onto Spotify is DistroKid's job.

We cannot reach a level the limiter cannot reach

At the loudest setting above, the measured result lands short. We print what came out, not what you asked for.

We cannot predict how far the encoder will move your peaks

It depends on the material, so the tool encodes and measures per track. Anyone quoting one number has not measured it.

We cannot tell you whether it sounds good

Everything here is a measurement, and taste is not one. The report says what changed and by how much; whether that is the record you wanted is yours to judge, which is why the free tier hands you the WAV instead of a watermarked preview.

Questions that come next

All 9, each with a measurement

Put your own track through it

The first three this month are free and the WAV comes with them, so the cheapest way to find out whether this is any good is to stop reading and try it on something you know well.

Master a trackCheck a mix first

How every number here was measured, what happens to your music, and what you would be buying.