Volume and level matching
Lift a recording that came out too quiet, or bring a set of files recorded at different levels into line. Peak (dBTP) and loudness (LUFS) are measured before and after, and any setting that would distort is flagged before you run it. Exports WAV and MP3.
Input
How to set the level
Each file is scaled so its loudest moment lands here, which means a different amount per file. −1 dBTP is the safe choice; 0 can distort once the file is re-encoded.
Holds the gain back if the peak would pass −1 dBTP. Turn it off and the full amount is applied, but anything above 0 dB is chopped flat on export and audibly distorts.
Export
WAV is 24-bit and uncompressed, so nothing is lost. MP3 re-encodes, so choose WAV if the file is going back into an editor. Output is two channels; a WAV source keeps its own sample rate, other formats come out at 44.1 kHz.
Output
Related tools
All tools- Audio cut & joinCut the part you need out of an audio file, or join several into one.
- Silence removerFind the silent stretches in a recording and close them up.
- Voice recorderRecord from your microphone, play it back, save as WAV or MP3.
- Tuner & metronomeMeasure pitch from the microphone in cents, with a metronome on the same page.
Highlights
- Measured, not predicted
- Both columns come from a full pass over the actual samples. The "after" figures are taken from the same gain-applied audio the export is encoded from, not from arithmetic on the "before" figures. Peak is 4× oversampled true peak, and loudness is the gated BS.1770 measurement streaming services use.
- The shape of the sound is untouched
- All this does is multiply by a number — no compression, no saturation, nothing added. The relationship between the quiet parts and the loud parts is exactly preserved, which is what you want for material that is going back into an editor.
- Distortion is flagged first
- Anything above 0 dB is chopped flat on export and cannot be recovered. If the settings would do that, the warning appears before the run, and by default the gain is held at −1 dBTP instead. Files that were held back are marked in the list.
How to use it
Add your audio
Click the box to choose files or drag them in; several at once is fine. Each one is measured as it loads, and the list fills in with its peak (dBTP) and loudness (LUFS).
Say what you want
Choose "change by dB", "match the peak" or "match the loudness", then set the figure. The clip guard is on by default. The list shows the amount that will be applied to each file before anything runs.
Apply and save
Press Apply and the result is measured again so the before and after figures sit side by side. Save one file from the download button on its row, or use "Download all as ZIP" for the set. WAV or MP3, your choice.
The numbers, and which tool to use
- How is this different from the mastering tool (/audio/master)?
- This multiplies by a number. It does not reshape anything, so a track whose peaks already reach the ceiling cannot be made louder here. The mastering page does the reshaping: EQ, dynamics and a limiter, to reach a streaming target. "This recording is too quiet" or "these files are at different levels" belongs here; "finish this track for release" belongs there.
- Peak or loudness?
- Peak looks only at the single loudest instant. Two files normalised to the same peak can still sound wildly different, because a dense mix carries far more energy than a solo voice. Loudness (LUFS) weights the whole file the way an ear does, so it is the one to match when several recordings will be heard together. For making a single recording as loud as it safely goes, peak is enough.
- Is dBTP the same as dBFS?
- No. dBFS is the highest sample value; dBTP is the highest level the waveform actually reaches between samples. A file can sit at −0.1 dBFS on every sample and still cross 0 dB in between, which is what makes an apparently clean file crackle after MP3 encoding. The figure shown here is true peak, found by 4× oversampling.
- Which loudness target?
- Use whatever the destination asked for. Failing that, podcasts and spoken audio usually sit near −16 LUFS and video platforms near −14 LUFS, while broadcast and quieter listening use −23 LUFS. If the source peaks are already near the ceiling, the clip guard will stop the gain short of the target rather than distort to reach it.
Questions
- Does this degrade the audio?
- The gain itself does not. The export format matters more: WAV is 24-bit and uncompressed, MP3 is a re-encode. Choose WAV if the file is going back into an editor. Note also that turning a recording down turns its noise floor down with it — and turning it up brings that noise up too.
- What happens if I turn the clip guard off?
- The full amount is applied and anything above 0 dB is flattened when the file is written. That is distortion, and it is not reversible from the exported file. Leave it on unless the distortion is what you are after.
- How many files at once?
- There is no fixed limit, but every decoded file is held in memory. Dozens of full-length tracks will make a modest machine struggle. Measurement also runs one file at a time, so the wait grows with the number and the length of what you add.
- What comes out?
- Two channels. A WAV source keeps its own sample rate; other formats come out at 44.1 kHz, because a compressed file does not reveal its rate until it has been decoded. A mono source is written as two identical channels, because some players mishandle a single-channel WAV. WAV output is 24-bit.
- Are my files uploaded?
- No. Decoding, measurement, processing and encoding all happen in the page, and closing the tab discards everything that was loaded.