Before you begin
  • 0 dBFS is the full-scale sample reference.
  • RMS and sample peaks describe different things.
  • Decoded sample rate is not always the original file’s sample rate.

Ask the meter a small, answerable question

A waveform can help you find a quiet introduction, see the overall duration, or compare broad level changes. It cannot tell you whether a performance is good, whether a mix is balanced, or whether you have permission to use the recording. Begin with a concrete question, such as “Is there a long silent section?” or “Did this export reach the sample ceiling?”

Understand the sample peak

The inspector finds the largest absolute decoded sample across channels and expresses it relative to full scale: 20 × log10(peak). A peak amplitude of 0.5 is about −6.02 dBFS; 1.0 is 0 dBFS. Digital silence has no finite logarithmic value, so the report shows negative infinity.

A peak near 0 dBFS means little sample headroom remains. It does not prove that clipping happened in the original recording. An already clipped signal could have been turned down later, while lossy decoding can create samples above the nominal range. This tool does not reconstruct that history.

RMS gives a different summary

RMS is the square root of the mean of squared samples. Here it is measured over all decoded samples and channels. A long quiet introduction lowers the overall RMS even if the chorus is loud. Peak and RMS together can help describe a file, but neither is an integrated LUFS reading or a judgement of perceived loudness.

Read “decoded” literally

The browser decodes the file through Web Audio. The displayed sample rate belongs to that decoded buffer, which may have been resampled to the audio context’s rate. Channel count also describes the decoded result. Do not cite these values as a complete technical inspection of the original container or codec metadata.

Know what the waveform leaves out

The display groups many samples into small time buckets and draws each bucket’s amplitude envelope. It is a navigation overview, not a sample-accurate editor. A dense-looking waveform does not establish a particular loudness, and a small waveform does not mean the recording has low fidelity.

A useful check after exporting

  1. Load the exported WAV or MP3 in Audio Inspector.
  2. Check its duration and listen for an unexpected ending or silence.
  3. Compare sample peaks and near-full-scale sample counts. If the mix sounds distorted, return to the source mix and lower its levels.
  4. Save the JSON report if you want a record of these decoded measurements.

For delivery requirements expressed as LUFS or true peak, use a meter designed for those measurements. The inspector intentionally reports only the simpler measurements it actually computes.

This guide describes the current product controls and implementation. Separation results depend on the recording, device and model; examples are not quality guarantees.

Further reading: Spleeter · Web Audio API · How the tools work