How to Test Audio Quality, Detect Clipping and Spot DC Offset
Quality here means technical health rather than artistic merit. Nothing on this page has an opinion about your mix. It measures four things that are objectively wrong when they are wrong, and each of them is easier to fix before publishing than after. Clipped samples are those sitting at or extremely close to full scale, counted across the file. A handful is normal and inaudible. Thousands mean the waveform has had its peaks flattened, either during recording at too high a gain or afterwards by processing, and that flattening adds harmonics that were never in the sound. Clipping cannot be undone, which is why catching it while you can still re-record matters. DC offset is a constant displacement of the whole waveform away from zero, usually from a faulty input or a badly behaved converter. You cannot hear it directly, but it eats headroom on one side of the signal, it can produce a click when audio starts or stops, and it interferes with anything downstream that assumes a centred signal. Silent lead-in is the gap before anything happens. Useful to know, because most platforms and listeners are unforgiving about a track that takes two seconds to begin. Spectral rolloff is the most informative of the four. It is the frequency above which there is effectively nothing, computed from an FFT across the file. Lossy encoders low-pass what they store, so a rolloff at 16 kHz means the audio was compressed at some point regardless of its current format. A file delivered as a lossless master, showing a 16 kHz cliff, was made from an MP3 and is not what it claims to be. Every one of these is measured from the decoded samples, so they describe the audio itself rather than what the container asserts about it.
How to Use the Audio369 Online Audio Quality Analyzer (Step-by-Step)
The whole thing is analysed rather than sampled, so a short burst of clipping in the middle is still counted.
A few is fine. Thousands means peaks have been flattened, and the only real fix is to go back to the source or re-record.
Offset should be near zero. A long silent lead-in is worth trimming before anyone else has to sit through it.
This is the file's history. A hard stop at 16 kHz means lossy encoding somewhere in its past, whatever the extension says now.
Technical Architecture & Audio Engine Specifications
What each reading means, and what to do
| Reading | Meaning | Fix |
|---|---|---|
| Thousands of clipped samples | ✓ Peaks flattened by too much gain | Re-record or return to the source; not repairable |
| DC offset well away from zero | ✓ The signal is not centred | A high-pass filter removes it |
| Rolloff at 15 to 16 kHz | ✓ Lossy encoding at a modest bitrate | Find a better source |
| Rolloff near 20 kHz | ✓ High-bitrate lossy or lossless | Nothing to do |
| Long silent lead-in | ✓ Dead air before the content | Trim it with the Audio Cutter |
Who Uses Audio369 Audio Quality Analyzer? (Practical Creative Workflows)
Checking a recording before you publish
Clipping and offset are invisible on a waveform at normal zoom and obvious in a measurement, which is the cheapest quality control there is.
Verifying a file you were sent
When a client or collaborator supplies a master, the rolloff tells you whether it is genuinely lossless or an MP3 in disguise.
Diagnosing distortion
If something sounds harsh but you cannot say why, a clipped sample count either confirms the cause or rules it out immediately.
Checking a new microphone or interface
DC offset is a hardware fault, and finding it early saves discovering it across a month of recordings.
Auditing an archive
Running old files through this shows which of them were transcoded from lossy sources at some forgotten point.
Frequently Asked Questions (FAQ)
Can clipping be repaired? +
Not really. The peaks were flattened, so the information about their shape is gone. Restoration tools guess at what was there, and a guess is not the recording. Re-record or go back to the source.
Is a small number of clipped samples a problem? +
No. A handful of samples at full scale across a whole file is inaudible. It is the difference between a few and thousands that tells you whether something went wrong.
How does rolloff prove a file was lossy? +
Lossy encoders discard everything above a cutoff to save bits, and that cutoff stays visible through any later conversion. A file with nothing above 16 kHz has been through an encoder, whatever its extension is now.
What causes DC offset? +
Usually a hardware fault in an input or a converter. It shifts the whole waveform off centre, which costs headroom and can click at the start and end. A high-pass filter removes it cleanly.
Does a good report mean my audio sounds good? +
No. This measures technical faults, not whether the performance, the balance or the room were any good. A file can pass every check here and still be a poor recording.
What is DC offset and why does it damage audio dynamics? +
DC offset occurs when audio waveforms center above or below the zero-voltage baseline due to faulty hardware or cheap audio interfaces. It robs headroom and causes clicks during editing cuts.
How does inter-sample clipping occur during lossy streaming playback? +
When consecutive digital samples reach 0 dBFS, digital-to-analog converters reconstruct curved analog peaks that overshoot full scale, causing harsh analog clipping distortion in consumer headphones.
Can the analyzer confirm if a high-res FLAC file is genuine? +
Yes. By measuring high-frequency spectral rolloff, the analyzer reveals whether a purported 24-bit 96 kHz FLAC file contains true ultrasonic harmonics or was simply upconverted from a low-res MP3.