Audio clipping detector
Open a recording to list every clipped spot and true-peak over with its time, then jump to each one and listen. Nothing is uploaded.
No file open yet
Drop an audio file here to find its clipped spots
wav, flac, mp3, m4a or ogg, 16-bit, 24-bit or float. The file is read on this device and never uploaded.
Reading the file…
0 %
Clipped spots
No file yet.
| Time | Ch | Length | What |
|---|
Press a time to jump there: the spot plays from 0.75 s before. ↑ ↓ move through the list, P plays the spot.
Detection
A spot counts as clipped when this many samples in a row sit at the same level at or above the threshold. Lower the threshold for a file that was clipped and then turned down.
File facts
- File
- Open a file to see its facts.
Export
The CSV lists every spot with its time, channel, length, flat-top runs, true peak and level.
Recent checks
How to use the clipping detector
- Press Open file or drop a file onto the tool. WAV, FLAC, MP3, M4A and Ogg all work, in 16-bit, 24-bit or 32-bit float.
- Wait for the scan. Every sample of every channel is checked, and a long WAV is read from the disk in parts with a progress bar.
- Read the four numbers under the charts: how many clipped spots there are, where the worst one sits, the sample peak and the true peak.
- Press any time in Clipped spots. The lower chart zooms into the samples there, and the spot plays from 0.75 seconds before, so you can hear whether the crackle is real.
- If needed, change Threshold or Minimum run under Detection, then save the list as a CSV for your editor.
How to read the list of clipped spots
Each row is one spot: a place where clipping happens, with its start time as minutes, seconds and milliseconds (00:01.400 is 1.4 seconds in), the channel (L or R) and how long it lasts. Clips less than 50 ms apart in one channel are grouped, so a hot snare hit that clips on eight wave peaks in a row is one spot of eight flat-top runs, not eight rows.
The What column names the kind. A flat top is a run of samples at the same level near full scale, the classic sign of digital clipping; the longest run is given in samples (at 48 kHz, 48 samples is 1 ms). True peak means the smooth wave between the samples goes over 0 dBTP. Over 0 dBFS only appears in 32-bit float files, which can store values above full scale.
The four readouts sum it all up. The worst spot is the one with the most clipped samples, and pressing P plays the chosen spot again.
Use cases
Podcast check
A 45-minute interview sounds fine until the guest laughs at 12:04 and the speaker crackles. Open the WAV and the list shows three spots in the guest’s channel, each 2 to 6 ms long. That is small enough for a de-clip tool to repair, and the times tell you exactly where to look instead of listening to the whole episode again.
Live recording
A gig recorded on a handheld recorder at 24-bit looks fine on its small meter. The detector finds 140 flat-top runs, all in the last song, when the drummer played harder and the input gain was too high. Every run sits on a kick drum hit. Now you know which song needs work and that the rest of the set is clean, and next time you can set the gain 6 dB lower.
Master before release
A dance track is mastered loud, about −8 LUFS. The list shows 60 flat tops of 3 to 5 samples and a true peak of +0.8 dBTP. Short flat tops like these are often a deliberate choice in loud music, but the true-peak over can turn into real distortion once the track becomes an MP3 or AAC for streaming. Lower the limiter’s ceiling to −1 dBTP and check the track again, then measure its LUFS and true peak.
Clipping vs true peak
A digital file stores the wave as samples, 44,100 or 48,000 points a second. Clipping happens to the samples: the wave wanted to go higher, but the samples could not, so they sit flat at the top. True peak is about what happens between the samples. When the file is played, the converter draws a smooth wave through the points, and that wave can bulge above the highest sample.
A tone at 12 kHz in a 48 kHz file is a clear example. With four samples per wave, the samples can land at 71 % of the real peak, so a file whose samples read −3 dBFS can hold a wave that peaks at 0 dBTP. The detector checks this with four times oversampling, the method that loudness standards use, and marks each place where the smooth wave passes 0 dBTP. A file can have no clipped samples at all and still have true-peak overs, which is why the summary gives both numbers.
Limits of the detector
Lossy files such as MP3 and AAC smear flat tops into small ripples, so a clipped MP3 may show true-peak overs instead of flat runs: check the WAV or FLAC master when you can. A master limited to a ceiling of −0.3 dBFS has flat-looking peaks but no runs at the same level, so it is not counted. Soft analog clipping in a preamp rounds the wave instead of cutting it flat, so it cannot be found this way, though your ears will still hear it.
How to fix clipping
The best fix is a new take with less gain: when you record at 24-bit, peaks around −12 to −6 dBFS leave plenty of room and add no noise you can hear. In a mix, lower the track or the master fader before the limiter, not after it. Light clipping of a few samples can be improved with a de-clip tool in an audio editor, which redraws the missing tops. Heavy clipping, with flat runs of many milliseconds, cannot be fully repaired. After the fix, set the limiter’s ceiling to −1 dBTP and scan the file again. For more checks before release, see the audio file checks, and make sure a lossless download is real with the fake FLAC checker.
FAQ
What is audio clipping?
When a signal is louder than the system can store, its peaks are cut flat, which sounds harsh or crackly.
What is true peak?
The real peak between samples after conversion back to analog. It can be over 0 dB even when samples are not.
Can clipping be repaired?
Light clipping can be improved with a de-clip tool in an editor; heavy clipping is best re-recorded.
Why does it find clips in a loud master?
Many loud masters are clipped on purpose a little; the list shows how often and how long.