Measure one audio track with FFmpeg volumedetect

The useful distinctions

• volumedetect takes no filter parameters.
• It reports statistics when the analyzed stream ends.
• The report includes mean and maximum sample levels.

Keep measurement separate from treatment

A level report answers a narrow numerical question about the selected recording. Decide which take and track you are measuring before considering a volume change. If a source contains both a microphone recording and a finished mix, measuring the wrong track can produce perfectly valid numbers for the wrong editorial decision. Keep the original recording intact throughout this first stage.

Wait for the complete pass

The example explicitly chooses one audio stream and produces no new listening copy. Preserve the diagnostic output after the analysis has finished so the result refers to the intended complete track. Comparing a short opening from one take with an entire second take creates an uneven comparison. Record any deliberate analysis boundary in the notes rather than presenting an excerpt’s result as though it described the whole recording.

Interpret the result carefully

A maximum sample value and an average measurement describe different aspects of the signal. Neither alone establishes intelligibility, listening comfort or compliance with a delivery specification. Use the report to frame the next question, such as why one take has a very different peak pattern, then inspect the relevant material. Avoid applying an automatic gain change merely because one number looks smaller. Any later processing should have its own purpose and verification.

A focused way to proceed

1. Select the intended audio track.
2. Run volumedetect through the full track to a null output.
3. Preserve the completed diagnostic output.
4. Compare like-for-like takes without treating the numbers as a complete perceptual assessment.

Worked example · synthetic

ffmpeg -i workshop-take.wav -map 0:a:0 -af volumedetect -f null -

Read the comparison

The report is sample-level evidence for the selected track; it is not a full perceptual loudness verdict.

When to stop or ask for help

If results look implausible, confirm the selected track and completed pass before requesting audio-specific advice.