How is the silence percentage worked out?

Every qualifying run is summed into a total silent duration, and that total is divided by the full decoded length of the file and shown to one decimal place. Because only runs that meet the minimum run length are counted, the figure is always a little lower than the raw amount of quiet in the recording, which is usually what you want when deciding how much can be cut.

What threshold should I start with?

The default of −60 dBFS is a reasonable first pass. Material that has been loudness normalized, such as a finished podcast, usually behaves sensibly between −50 and −60 dBFS, while a noisy room recording needs a less negative threshold so the noise floor does not register as programme material.

Why does a gap I can clearly hear not appear in the table?

Two settings can hide it. The gap may be shorter than the minimum run length, in which case it is discarded before any total is computed, or its frame RMS may sit above the threshold you chose. Lower the threshold and shorten the minimum run to see how the run count responds.

Does the tool remove the silence it finds?

No. The processing step is a pass-through: the analysis runs, the statistics footer fills in and the audio comes back untouched under a silence-stats suffix. Removing the gaps is a separate editing job, and the run table exists to give you the timestamps that job needs.

What does the minimum run setting change?

It sets the shortest gap that is counted at all. Runs under that length are ignored, so raising it filters out breaths and clicks while lowering it catches very short pauses. It is measured in milliseconds from 100 to 2000 in steps of 50, and it affects the run count, the silent time and the silence percentage together.

Silence Statistics Calculator

You Might Also Need

Silence Remover

Automatically remove silent sections

Audio Gate

Noise gate to silence quiet parts

Waveform Visualizer

Visualize audio waveforms in real-time

Audio Trimmer

Trim audio files to keep only the parts you need

Silence Statistics Calculator loads an audio file and reports how much of it is silence, how many separate silent stretches there are and how long each one runs. The scan is level based: you set a dBFS threshold and a minimum length, and every qualifying gap is listed with the timestamps you would need to cut it.

What is Silence Statistics Calculator?

Drop an audio file onto the panel and the tool decodes it once, scans the whole thing in fixed length blocks and fills in a results footer instead of processing the audio. The download button still works and returns your original file under a silence-stats suffix, but not one sample is changed: this tool reports on silence, it never cuts it out. Silence detection here is driven by two sliders and the numbers refresh as soon as either one moves.

The controls and the readout:

A silence threshold in dBFS, adjustable from −80 to −20 and defaulting to −60.
A minimum run in milliseconds, adjustable from 100 to 2000 in 50 millisecond steps and defaulting to 500.
A summary strip showing silent time, the silence percentage, the number of runs and the longest single run.
A scrollable table listing every qualifying run as a numbered row with start, end and length columns in seconds.
A pass-through download of the same audio, named with a silence-stats suffix.

Because the analysis re-runs on every slider move, the loop is quick: nudge the threshold, watch the run count change and stop when the table matches what you can hear. Two figures are worth understanding before you start. Silent time counts qualifying runs only, so it is always a little lower than the raw amount of quiet in the file, and the silence percentage divides that total by the full decoded duration, leader and tail included.

Under the hood the scan divides the signal into 20 millisecond frames. For each frame it sums the squared samples across every channel, takes the square root to get an RMS level in the linear domain and converts that to dBFS with the 20 log10 relation. A frame counts as silent when its RMS sits below the threshold. Consecutive silent frames are grouped into a run, and a run is only kept when its duration reaches the minimum run length; anything shorter is discarded before any total is computed. The surviving runs are summed for the silent time, averaged for the mean run and compared for the longest, so every figure on screen comes from the same list of qualifying gaps.

What the measurement genuinely cannot tell you:

Gaps shorter than the minimum run are dropped entirely, so the silence percentage never counts brief breaths, clicks or micro-pauses.
The threshold tests frame RMS rather than sample peak, so a low level passage can read as silence even when room tone is clearly audible.
A steady hum sitting above the threshold hides real silence and can make a very quiet recording report almost no silent time.
The scan cannot separate speech from music, so an intentionally quiet instrumental section is measured exactly like dead air.
Run boundaries land on the 20 millisecond frame grid, so start and end times are accurate to that resolution rather than to the sample.
It only reports; there is no automatic removal and no way to export the silent regions on their own.

A threshold that suits one file will not suit the next, so slide it once before you trust the numbers. Podcast episodes that have already been loudness normalized usually read sensibly between −50 and −60 dBFS, while a noisy room recording needs a less negative figure so its noise floor is not mistaken for programme material. The table is the part to use: those start and end times are exactly the values a silence remover expects, so measure here first, note the runs worth cutting and only then edit.

How to use Silence Statistics Calculator

Drop an audio file onto the panel or click to browse; the file is decoded once and the statistics footer appears.
Slide the silence threshold until quiet passages are being counted the way you would count them by ear.
Set the minimum run length so short breaths and clicks are excluded and only meaningful gaps remain.
Read the summary and the run table, then copy the start and end times of the runs you intend to remove.

When to use Silence Statistics Calculator vs related tools

Silence Statistics Calculator answers a question before anything is edited, so reach for it when you need to size a recording up rather than change it. Silence Remover is the tool that acts on what you find here, dropping the gaps outright with the same level based logic, while Audio Gate suppresses quiet passages in place instead of deleting time and suits continuous material such as a noisy interview. When a long gap also needs a tidy cut, Audio Trimmer does the trimming, and Waveform Visualizer gives you the picture to check boundaries against before committing.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.