The last check before you release or play out: every reading, what to fix first, a set check for DJs, reference compare, mastering, stems and a vocal remover.
Drop up to 50 tracks. Each gets its loudness, its true peak, how dynamic it is, the gain trim that levels it with the rest, and one line on the file itself. Checked on your own computer; nothing is uploaded. The lead of the DJ plan.
DJ and StudioEvery reading, with what to fix first
DC offset, silence, real bit depth, a cut-off ending and the sample rate, and what no meter shows you: a WAV or FLAC that was an MP3 or Opus file before.
Hold any reading against the meter you trust, and write to us if one disagrees. Only Spotify and SoundCloud publish a loudness target, and the report says where every other platform’s figure comes from.
Integrated loudness−9.5 LUFSLoud, by choiceTrue peak+0.61 dBTPOutsideCorrelation0.87In band
03
Read every reading
Each one with its target band and a verdict.
Fix firstTrue peak is 2.61 dB over the ceiling
04
Fix what comes first
The one change that matters most, at the top.
ABDrop against dropLevel-matched by −1.97 dB
05
Hold it against a reference
Your sections against its sections, at matched loudness.
app.refbooth.com/s/••••••••Link copied
06
Share it with a link
Anyone with the link reads it, no account needed.
Meet RefBooth, the second opinion on every master.
Every figure here is what RefBooth measured in an example track made for this page, drawn with the instruments of the app.
Loudness
Loud, and how loud, from the first bar to the last.
Integrated loudness, its range and its maxima, and the short-term curve over the whole track, marked by section and set against the target of the platform you deliver to.
−9.5−9.5 LUFSintegrated, against Spotify’s −14
LUFS · short-term 3 s at 10 HzShort-termIntegratedTarget SpotifyGated outTrue-peak overs
True peak
The peak after every codec, not only in your file.
Your master as delivered, then again through the formats the platforms stream in: Spotify at 96, 160 and 320 kbps, YouTube and TIDAL. An over that only appears after encoding is found before release.
+0.61+0.61 dBTPin the file as delivered
TP max · Whole track
0.61dBTP
0.91 dB over the ceiling
+0.61 dBTP, 0.91 dB over the −0.3 ceiling, 4 ISPs
After the delivery encodeScaleDecoded dBTPΔ vs sourceISP
Source as uploaded+0.61over—4
+60−6−12−18−24dBTP · ceiling −0.30
Spotify Ogg Vorbis 96+0.01over−0.601
Spotify Ogg Vorbis 160−0.08over−0.690
Spotify Ogg Vorbis 320+0.57over−0.035
YouTube Opus 160+0.37over−0.246
TIDAL FLAC+0.61over0.004
Apple Music AAC 256 — Not measured with the platform’s encoder
YouTube AAC-LC 128 — Not measured with the platform’s encoder
SoundCloud AAC 256 · 160 · 96 — Not measured with the platform’s encoder
TIDAL AAC 320 — Not measured with the platform’s encoder
ISP
4
Clipped samples
0
True peak at
1:47.7
TP max · Whole track
0.61dBTP
0.91 dB over the ceiling
+0.61 dBTP, 0.91 dB over the −0.3 ceiling, 4 ISPs
After the delivery encodeDecoded dBTPΔ vs sourceISP
Source as uploaded+0.61over—4
0−12−24dBTP · ceiling −0.30
Spotify Ogg Vorbis 96+0.01over−0.601
Spotify Ogg Vorbis 160−0.08over−0.690
Spotify Ogg Vorbis 320+0.57over−0.035
YouTube Opus 160+0.37over−0.246
TIDAL FLAC+0.61over0.004
Apple Music AAC 256 — Not measured with the platform’s encoder
YouTube AAC-LC 128 — Not measured with the platform’s encoder
SoundCloud AAC 256 · 160 · 96 — Not measured with the platform’s encoder
TIDAL AAC 320 — Not measured with the platform’s encoder
ISP
4
Clipped samples
0
True peak at
1:47.7
TP max · Whole track
0.61dBTP
0.91 dB over the ceiling
+0.61 dBTP, 0.91 dB over the −0.3 ceiling, 4 ISPs
After the delivery encodeDecoded dBTPΔ vs sourceISP
Source as uploaded+0.61over—4
0−12−24dBTP · ceiling −0.30
Spotify Ogg Vorbis 96+0.01over−0.601
Spotify Ogg Vorbis 160−0.08over−0.690
Spotify Ogg Vorbis 320+0.57over−0.035
YouTube Opus 160+0.37over−0.246
TIDAL FLAC+0.61over0.004
Apple Music AAC 256 — Not measured with the platform’s encoder
YouTube AAC-LC 128 — Not measured with the platform’s encoder
SoundCloud AAC 256 · 160 · 96 — Not measured with the platform’s encoder
TIDAL AAC 320 — Not measured with the platform’s encoder
ISP
4
Clipped samples
0
True peak at
1:47.7
Spectrum
Where the energy sits, band by band.
The whole track’s spectrum from the sub to the air, with each named region read out, so a heavy low end or a dull top shows before anyone else hears it.
Spectrum
Long-term average of the whole track, from the report
dB · 1/6 oct · slope 4.5 dB/oct around 1 kHz · Mid (L+R)
Long-term average
Max hold
Spectrum
Long-term average of the whole track, from the report
dB · 1/6 oct · slope 4.5 dB/oct around 1 kHz · Mid (L+R)
Long-term average
Max hold
Spectrum
Long-term average of the whole track, from the report
dB · 1/6 oct · slope 4.5 dB/oct around 1 kHz · Mid (L+R)
Long-term average
Max hold
Stereo width
Width you can see, in every band.
How wide each band sits against your reference, with correlation and balance, so a wide top and a mono-safe low end are a reading, not a guess.
0.870.87correlation, the whole track
side share · √ scale · 1/3 octave, whole trackYoursReference
Width
0.063
whole track · 0 = mono
Correlation
0.874
whole track
Bass correlation
0.999
below 120 Hz
Mono-sum loss
−0.28dB
side share · √ scale · 1/3 octave, whole trackYoursReference
Width
0.063
whole track · 0 = mono
Correlation
0.874
whole track
Bass correlation
0.999
below 120 Hz
Mono-sum loss
−0.28dB
side share · √ scale · 1/3 octave, whole trackYoursReference
Width
0.063
whole track · 0 = mono
Correlation
0.874
whole track
Bass correlation
0.999
below 120 Hz
Mono-sum loss
−0.28dB
Compare
Drop against drop, not minute against minute.
Add a reference and RefBooth pairs your sections with its sections, in order, so two drops meet two drops, and compares each pair at matched loudness, so a louder chorus never passes for a better one. Compare is part of Studio.
44section pairs, an example master against its reference
Both tracks were made for this page, so no one else’s music is shown.
Difference, drop against drop
Yours against Reference, band by band
dB · yours − Reference after level match · 1/6 oct · Drop ↔ Drop
Yours − Reference
±3 dB
Outside
Level match−1.97dBto yours, by section
Difference, drop against drop
Yours against Reference, band by band
dB · yours − Reference after level match · 1/6 oct · Drop ↔ Drop
Yours − Reference
±3 dB
Outside
Level match−1.97dBto yours, by section
Difference, drop against drop
Yours against Reference, band by band
dB · yours − Reference after level match · 1/6 oct · Drop ↔ Drop
Yours − Reference
±3 dB
Outside
Level match−1.97dBto yours, by section
From the first bar to the last codec, every reading of your master sits against the platform you deliver to. What to fix first comes ranked, and the reference you trust sits beside it at matched loudness. Everything a finished master needs checked, in one report.
Every reading, against its target. 11 of 13 in band against Spotify.
An example master against Spotify: 11 of 13 readings in band
Reading
Value
Target band
Verdict
Loudness
Integrated loudness
−9.5 LUFS
−16.0 to −13.0 LUFS
Loud, by choice
Loudness range
4.5 LU
4.0 to 12.0 LU
In band
Short-term max
−7.8 LUFS
−24.0 to −11.0 LUFS
Loud, by choice
Peaks
True peak
+0.61 dBTP
−6.00 to −2.00 dBTP
Outside
Inter-sample peaks
4
zero
Outside
Sample peak
−1.00 dBFS
−6.00 to −0.10 dBFS
In band
An example master against Spotify: 11 of 13 readings in band, continued
Reading
Value
Target band
Verdict
Dynamics
Peak to loudness ratio
10.1 LU
9.0 to 18.0 LU
In band
Crest (TP − RMS)
10.5 dB
10.0 to 20.0 dB
In band
Level variation
5.3 dB
3.0 to 10.0 dB
In band
Stereo
Correlation
0.87
0.30 to 1.00
In band
Bass correlation
1.00
0.70 to 1.00
In band
Mono-sum loss
−0.28 dB
−1.00 to 0.00 dB
In band
Spectrum
Spectral slope
−3.5 dB/oct
−3.6 to −1.8 dB/oct
In band
What to fix first. The findings, ranked, each with what to do about it.
Fix firstTrue peak is 2.61 dB over the ceiling
+0.61 dBTP against the Spotify ceiling, −2.0 dBTP, worst at 1:48. 4 inter-sample peaks (ISP) above 0 dBTP. Inter-sample peaks appear when a master was not limited with an oversampling (true-peak) limiter. A converter rebuilds them in its analog stage, up to about 3 dB past full scale, and a lossy encoder clips them.
FixCorrelation dips to −0.365 at 1:21
For 400 ms (Breakdown) the left and right channels pull against each other: what plays there partly cancels on a mono system, a club sub summed to mono, a phone. The whole-track correlation, 0.87, averages it away. Jump there and listen in mono to hear what goes.
NoteLoud by choice: 3.5 LU above the range
−9.5 LUFS against the Spotify range, −16.0 to −13.0 LUFS. Spotify turns it down 4.5 dB on playback; the density used to reach it stays. Loud genres make this trade on purpose; what it cost shows in the dynamics readings.
Sections, paired. Your sections against the reference’s, in order, each pair with its loudness, range and peaks. Compare is part of Studio.
Sections
Your sections paired with Reference’s
Yours
Bars
Reference
Pairing
Loudness LUFSyours · ref · Δ
vs loudest dByours · ref · Δ
Integrated LUFSyours · ref · Δ
Short-term max LUFSyours · ref · Δ
True peak dBTPyours · ref · Δ
PSR min LUyours · ref · Δ
Groove 10:00–1:17
1–40
Groove 10:00–1:17
same label
−9.1 · −7.1Δ −2.0 LU
0.0 · 0.0Δ 0.0 dB
−9.0 · −7.0Δ −2.0 LU
−8.3 · −6.3Δ −2.0 LU
0.30 · 1.04Δ −0.74 dB
7.6 · 7.3Δ +0.4 LU
Breakdown1:17–1:48
41–56
Breakdown1:17–1:48
same label
−13.0 · −10.3Δ −2.8 LU
−4.0 · −3.2Δ −0.8 dB
−12.8 · −10.1Δ −2.8 LU
−11.5 · −8.8Δ −2.7 LU
0.61 · 1.96Δ −1.35 dB
7.5 · 6.6Δ +0.8 LU
Drop1:48–2:35
57–80
Drop1:48–2:35
same label
−9.1 · −7.1Δ −2.0 LU
0.0 · −0.1Δ 0.0 dB
−8.9 · −7.0Δ −2.0 LU
−7.8 · −5.8Δ −2.0 LU
0.09 · 1.47Δ −1.38 dB
7.7 · 7.3Δ +0.4 LU
Groove 22:35–2:50
81–88
Groove 22:35–2:50
same label
−9.8 · −7.9Δ −2.0 LU
−0.8 · −0.8Δ 0.0 dB
−10.2 · −8.2Δ −2.0 LU
−9.5 · −7.5Δ −1.9 LU
−0.88 · 0.81Δ −1.69 dB
8.6 · 8.4Δ +0.2 LU
Sections
Your sections paired with Reference’s
Yours
Reference
Loudness LUFSyours · ref · Δ
vs loudest dByours · ref · Δ
Integrated LUFSyours · ref · Δ
Short-term max LUFSyours · ref · Δ
True peak dBTPyours · ref · Δ
PSR min LUyours · ref · Δ
Groove 10:00–1:17
Groove 10:00–1:17
−9.1 · −7.1Δ −2.0 LU
0.0 · 0.0Δ 0.0 dB
−9.0 · −7.0Δ −2.0 LU
−8.3 · −6.3Δ −2.0 LU
0.30 · 1.04Δ −0.74 dB
7.6 · 7.3Δ +0.4 LU
Breakdown1:17–1:48
Breakdown1:17–1:48
−13.0 · −10.3Δ −2.8 LU
−4.0 · −3.2Δ −0.8 dB
−12.8 · −10.1Δ −2.8 LU
−11.5 · −8.8Δ −2.7 LU
0.61 · 1.96Δ −1.35 dB
7.5 · 6.6Δ +0.8 LU
Drop1:48–2:35
Drop1:48–2:35
−9.1 · −7.1Δ −2.0 LU
0.0 · −0.1Δ 0.0 dB
−8.9 · −7.0Δ −2.0 LU
−7.8 · −5.8Δ −2.0 LU
0.09 · 1.47Δ −1.38 dB
7.7 · 7.3Δ +0.4 LU
Groove 22:35–2:50
Groove 22:35–2:50
−9.8 · −7.9Δ −2.0 LU
−0.8 · −0.8Δ 0.0 dB
−10.2 · −8.2Δ −2.0 LU
−9.5 · −7.5Δ −1.9 LU
−0.88 · 0.81Δ −1.69 dB
8.6 · 8.4Δ +0.2 LU
What every platform does with it. Each target, where it comes from, and what that platform does with the example master. Sources for every figure
Spotify−14 LUFSPublished by the platformTurned down 4.5 dBOver the −2.0 dBTP ceiling
Apple Music−16 LUFSPartly published by the platformTurned down 6.5 dBOver the −1.0 dBTP ceiling
YouTube−14 LUFSReported, not publishedTurned down 4.5 dBOver the −1.0 dBTP ceiling
SoundCloud−14 LUFSPublished by the platformTurned down 4.5 dBOver the −2.0 dBTP ceiling
TIDAL−14 LUFSReported, not publishedTurned down 4.5 dBOver the −1.0 dBTP ceiling
Amazon Music−14 LUFSReported, not publishedTurned down 4.5 dBOver the −2.0 dBTP ceiling
Deezer−15 LUFSReported, not publishedTurned down 5.5 dBOver the −1.0 dBTP ceiling
Club / DJNot levelledA working conventionPlayed as isOver the −0.3 dBTP ceiling
Read fromthe ground up
LoudnessIntegrated loudness · Loudness range · Short-term max
PeaksTrue peak · Inter-sample peaks · Sample peak
DynamicsPeak to loudness ratio · Crest (TP − RMS) · Level variation
StereoCorrelation · Bass correlation · Mono-sum loss
SpectrumSpectral slope
SectionsGroove 1 · Breakdown · Drop · Groove 2
What to fix3 findings, ranked
Open the app
The analyzer lives at app.refbooth.com, in any browser: your library, every report, and Compare beside it, with the instruments arranged the way you work.
A picture of the RefBooth analyzer with an example master open against Spotify: loudness, loudness history, true peak after each streaming codec, spectrum and width by band.
Example master2:51WAV · 48 kHz · 24-bitAgainst Spotify
LoudnessIntegrated, range, maxima
Target Spotify −14.0 LUFS
IL · Integrated
−9.5LUFS
above the Spotify target
LRA
4.5LU
SL max
−7.8LUFS
ML max
−7.3LUFS
TP max
0.61dBTP
ceiling −1.00
PLR
10.1LU
Loudness historyShort-term, the whole track
LUFS · short-term 3 s at 10 HzShort-termIntegratedTarget SpotifyGated outTrue-peak overs
True peakYour file, then after each platform’s codec
TP max · Whole track
0.61dBTP
0.91 dB over the ceiling
+0.61 dBTP, 0.91 dB over the −0.3 ceiling, 4 ISPs
After the delivery encodeDecoded dBTPΔ vs sourceISP
Source as uploaded+0.61over—4
0−12−24dBTP · ceiling −0.30
Spotify Ogg Vorbis 96+0.01over−0.601
Spotify Ogg Vorbis 160−0.08over−0.690
Spotify Ogg Vorbis 320+0.57over−0.035
YouTube Opus 160+0.37over−0.246
TIDAL FLAC+0.61over0.004
Apple Music AAC 256 — Not measured with the platform’s encoder
YouTube AAC-LC 128 — Not measured with the platform’s encoder
SoundCloud AAC 256 · 160 · 96 — Not measured with the platform’s encoder
TIDAL AAC 320 — Not measured with the platform’s encoder
ISP
4
Clipped samples
0
True peak at
1:47.7
Spectrum
Long-term average of the whole track, from the report
dB · 1/6 oct · slope 4.5 dB/oct around 1 kHz · Mid (L+R)
Long-term average
Max hold
Width by bandAgainst Reference
side share · √ scale · 1/3 octave, whole trackYoursReference
WAV, AIFF, FLAC, MP3, M4A, AACUp to 400 MB and twenty minutes
Loudness
Integrated loudness LUFS: how loud the whole track sounds, in the unit streaming platforms set their targets in.Loudness range LRA: how far the loudness moves between the quiet parts of the track and the loud ones, in LU.Short-term max
Peaks
True peak dBTP: the highest level the track reaches when it is played, between the samples too. Platforms give a ceiling in it.Inter-sample peaksSample peak
Dynamics
Peak to loudness ratioCrest (TP − RMS)Level variation
Stereo
CorrelationBass correlationMono-sum loss
Spectrum
Spectral slope
Platforms
Spotify, Apple Music, YouTube, SoundCloud, TIDAL, Amazon Music, Deezer, Club / DJEach target with where it comes from
Report
Every reading against its target bandWhat to fix first, rankedSections, by name, across the track
File checks
Format, lossy source, bit depth, fake stereoDC offset, clipped runs, dead air, a chopped end
Trial
3 days on your own tracksCancel before they end and nothing is charged
DJ
Set check for DJs: up to 50 tracks side by side, with the gain trim for eachStem separation: vocals, drums, bass and chords/synthsVocal remover: the instrumental and the acapella
Studio
Compare up to 8 sources at once, level-matchedLevel-matched A/B player with band soloOnline mastering for any platform or to your own figures, in three styles and three strengths
Pricing
Two plans, DJ and Studio, each with a 3-day trial. Cancel in those three days and nothing is charged.
DJ
For DJs, and for anyone who needs the report: every track checked, your set levelled, stems and the vocal remover.
$9a month
Or $79 a year.
Sales tax is added at checkout where it applies. Renews automatically until you cancel.
Online mastering for any platform or to your own figures, in three styles and three strengths
The mastering assistant: describe the master in your words, then check it against what was measured
Simulate fix: a peak fix rendered and measured before you make it
Album mode (24 tracks) and comments
100 share links
500 tracks, 50 GB
RefBooth cannot be bought from the country selected.
Both start with a 3-day trial. Your card is charged when the three days end; cancel before then and nothing is charged. Switch between DJ and Studio at any time. First charge refundable within 14 days. Cancel any time.
Prices are in euros in the EEA, in pounds in the UK and in US dollars everywhere else. Pick the country you are billed in: its price is the one that applies, as the terms set out.
Questions
Before you drop a master.
Is there a free plan?
No. There are two plans, DJ and Studio, and each starts with a 3-day trial. Cancel in those three days and nothing is charged.
Is RefBooth for DJs who don’t produce?
Yes. The DJ plan checks the tracks you play. Set check puts up to 50 side by side with the trim for each, and the quality line tells you which file is clipped, cut short or not what it claims to be. No studio and no DAW needed.
My DJ software already sets the gain. Why check?
Auto gain sets a level as a track loads. Set check shows the trim for every track before the gig, with its peak and the state of the file, so you know which tracks will push you into the red and which file to replace.
Do I need a card for the trial?
Yes. Stripe holds it, and nothing is charged until the three days end. We write to you the day before.
Can I see a report before I upload anything?
Yes. The sample report is a real one, on a real master, with nothing to sign up for: app.refbooth.com/sample.
Which files can I drop?
WAV, AIFF, FLAC, MP3, M4A and AAC, up to 400 MB and twenty minutes. Drop the file you are about to deliver: the report reads it as it is.
What happens to my master?
It is stored and measured on our own servers in the European Union, shared with no other company and never used to train anything. You can delete it at any time, and deleting it removes the audio and its report. A master uploaded without an account is deleted after 7 days.
How is this different from a loudness meter?
A meter shows the moment that is playing. RefBooth reads the whole file, judges every reading against the platform you deliver to, and ranks what to fix first, and shows what each streaming format does to your peak. Studio adds a reference, compared section by section at matched loudness.
Can RefBooth master my track, or split it into stems?
Yes. Mastering is part of Studio: it prepares your track for a platform, at the target it publishes or −14 LUFS where it publishes none, or to your own figures, saved as a new version with its own report and an A/B against the original. Stem separation and the vocal remover are in both plans: vocals, drums, bass and chords/synths in a mixer, or the instrumental and the acapella, to download.
Where do the platform targets come from?
From each platform where it publishes one. Where it does not, the report says so: Spotify and SoundCloud publish a target, Apple publishes a headroom rule, and the rest are reported figures or working conventions, marked as such beside the number.
Can I cancel, and can I get my money back?
Yes. Cancel in the first three days and nothing is charged. After that you can cancel at any time, and your plan keeps working until the period you paid for ends. The first charge is refundable within 14 days, no questions asked, from Settings in one click.
Why does the price depend on my country?
Prices are set by country so that they are fair where you live: in euros with VAT included across the European Economic Area, in pounds with VAT included in the United Kingdom, and in US dollars everywhere else, at five price levels. We suggest a country from your connection and you can pick another. Pick the country you are billed in: its price is the one that applies, as the terms set out.
Does this site store anything on my device?
Only what you choose: the country you pick for prices, until you close the browser, and your answer about visit statistics. Dropping a file into the box also sets one cookie for the app at app.refbooth.com, so the report is yours when you come back (seven days, or until you sign in). Statistics are counted by our own tool on our own server, set no cookie and are never linked to an account.
Write to the people who build RefBooth
Every message lands with the founders, not a ticket queue. Pick what it is about, or just write.
