Define the preparation problem
A sequence of authorised clips can force the audience to keep adjusting volume even when each file plays correctly. This is an authoring problem about listening consistency, separate from a muted player or wrong output device. The loudest sample peak and the perceived level across a programme are not the same measurement. A clip with one sharp transient can have a high peak while most speech remains quiet.
Before processing, identify the delivery specification and audience. Broadcast, online training and a local presentation need not use the same required targets. EBU R128 is a documented loudness framework, while FFmpeg's loudnorm filter implements EBU R128 loudness normalisation with several modes. Do not choose an arbitrary target merely because it appears in a command copied from a forum.
A hypothetical two-clip sequence
Imagine a creator combines a quiet interview and a loud promotional introduction that they own. Raising the computer's volume for the interview makes the next clip uncomfortable. Changing the device slider during review does not change the files delivered to the audience. The creator needs a controlled preparation and review process across the sequence, not repeated manual compensation at playback time.
Listen at a stable, comfortable reference level and measure the complete relevant programmes through the chosen tool. Include the entire intended clip rather than a convenient loud sentence. Keep the unprocessed sources. If a clip contains deliberate quiet passages, the creative intent matters; consistency should not mean forcing every moment to the same instantaneous level.
Keep measurement and processing settings traceable
FFmpeg documents single-pass and double-pass workflows for loudnorm, with targets involving integrated loudness, loudness range and true peak. A measured-file workflow should retain the analysis associated with the exact source version. Reusing measurements from an earlier edit can invalidate the intended processing assumptions. Use the installed tool's documentation for its actual options and reported output.
Loudness normalisation is not a repair for every intelligibility issue. Background noise, competing music, distortion or a poor recording may require separate editorial work. Avoid promising that a numeric target will make all speech clear. Evaluate the result through the intended speakers or headphones, while protecting your hearing with a comfortable playback level.
Review transitions and the complete deliverable
Compare the transition between clips as well as each clip in isolation. Confirm that peaks remain within the required specification and that no obvious pumping or distortion has appeared. Check audio synchronisation after export. A measurement report is useful, but the audience experiences a sequence, so the final review should follow that sequence.
Document the source versions, chosen delivery target, analysis and processing method. State that the result was prepared for the specified destination rather than claiming one universal best loudness for every service. A successful outcome is a consistent authorised programme whose technical measurements and listening review agree with the task, while the original recordings remain available for future revisions.
Questions answered
Is the player's volume setting a file loudness adjustment?
No. It changes that listening session. Preparing a deliverable requires an intentional authoring and measurement workflow.
Should every video use the same target from an online example?
Use the destination's actual specification and review the content. Different delivery contexts can have different requirements.
Sources & further reading
Official references for the subjects discussed. Service terms and device software can change.