Skip to content

Why your DAW meter and Sonalix disagree

Send feedback

Last updated 3 min readTags:AnalysisMetering

Compare the exact export, meter mode, time span and tolerance before interpreting a difference.

Reimport the exact file you uploaded. Play it through a clean path at unity gain, with no master processing or auto-normalisation. Reset the meter, play the file from beginning to end, and compare the matching reading. This rules out most differences caused by the export or by a meter watching only part of the track.

Match the measurement

Report label DAW reading to compare
Overall loudness Integrated loudness in LUFS
Highest peak True peak in dBTP
Highest sample Sample peak in dBFS
Peaks above the average PLR, calculated as true peak minus integrated loudness
Quiet-to-loud variation Loudness range (LRA) in LU
Left-right similarity Stereo correlation
Level change in mono Gain change from the stereo signal to its mono fold-down, using the same convention

Sonalix decodes the file to 32-bit float at its original rate and layout. If you used export normalisation, sample-rate conversion, dither or a lossy codec, the report measures the result of those steps. With a lossy upload, peaks and high-frequency content reflect the decoder's output. Compare a local meter on that same file. A DAW's RMS or crest-factor reading measures something different. The report does not compute crest factor. See true peak.

Check the span and the gates

Integrated loudness covers the whole file. Momentary loudness uses 0.4 s windows, and short-term loudness uses 3 s windows. Sonalix evaluates maxima every 100 ms, so the momentary maximum can read up to 0.58 LU lower than one measured on a finer time grid. See integrated, short-term and momentary loudness.

Integrated loudness uses a −70 LUFS absolute gate and a second gate 10 LU below the loudness calculated after the first gate. An ungated meter can read lower if the track contains long quiet stretches. Sonalix calculates LRA using complete 3 s windows and gates of −70 LUFS and −20 LU. It withholds LRA below five gated windows and marks it provisional for files under 60 s.

Check true-peak tolerance and mono routing

  • True peak. Sonalix displays 0.1 dB increments. Its readings on the EBU validation signals were within +0.2/−0.4 dB of the reference. A peak in the first 100 ms can cause an over-read of about 0.5 dB. A difference of a few tenths near full scale is plausible. Check both meters' settings before treating either as definitive.
  • Sample rate. True peak is validated at 44.1 and 48 kHz. Sonalix marks 32, 88.2, 96 and 176.4 kHz provisional. Other rates have no true-peak value.
  • Mono routing. Sonalix measures a one-channel file as one channel and also calculates a dual-mono loudness for PLR. Feeding the same signal at unity to two channels adds 3.01 LU to the meter reading. Check your DAW's routing and pan law before making that comparison. Sonalix also reports true peak at or above sample peak.

Compare charts with the same averaging

Sonalix's spectrum uses a Hann window, a hop of one quarter of the window and a bin width of at most 1.465 Hz. It averages linear power over the whole file, including silence. A live spectrum usually shows a shorter window with its own smoothing. The shapes may agree even when the values differ.

Whole-track correlation is measured at zero lag without removing the mean. Its time series uses 3 s windows. A shorter live window can swing more. The tonal findings use active seconds within 4 dB of the loudest 20 s and compare them with a reference balance. A live analyser has no equivalent unless it uses the same selection and reference.

If a sizeable gap remains after these checks, meter the same export with a second tool. For exact details, see measurement methodology and your mix report. Spotify's −14 LUFS Normal setting describes playback gain. Compare the unadjusted local files when checking the meters.

Was this page helpful?

    Type to search titles, headings and summaries.