My ad audio never clips. Why does it still sound much louder than other videos?

Separate peak safety from listening experience

A voiceover can be consistently forceful even when its peaks stay below the ceiling. Conversely, one short impact can have a high peak while most of the ad sounds quiet. Record both observations rather than treating a green peak meter as approval of the entire soundtrack.

Compare the ad with a relevant reference at the same player volume and device setting. Avoid comparing a phone speaker with headphones or one player’s normalized playback with another player’s raw file. Keep the reference identity and playback conditions so the comparison remains useful rather than a vague loud or soft judgement.

Measure the finished programme from beginning to end

Adobe distinguishes integrated whole-program, short-term and momentary loudness readings and states that its meter does not alter the signal. Record the tool/meter mode, whole-file integrated reading with its displayed unit, and whether the peak reading is sample peak or true peak rather than treating every peak meter as identical. ITU-R BS.1770 separately addresses programme loudness and true-peak measurement; it is not a universal target for every publishing platform. Reset measurement, play the complete final file without skipping, retain the actual delivery requirement and listen to speech under the same playback conditions.

Use a target only when the actual publisher or delivery specification provides it. If none is specified, record the evidence and seek a production decision without claiming compliance. The selected adjustment must preserve understandable speech and avoid unnecessary pumping, distortion, or a sudden loud ending.

1. Fix the listening device and player volume.
2. Reset an external loudness meter for the final file.
3. Measure the complete advertisement once without skipping.
4. Record peak, integrated reading, and distracting sections.
5. Adjust externally and repeat the same full-file comparison.

A fictional short-ad audio review

Fictional exercise: a fifteen-second ad has a dense music bed and an energetic voiceover. Its peaks do not clip, but it feels louder than the chosen comparison. The review records the whole-file loudness and the sections that obscure speech. No actual CartVisual audio output or platform result was measured.

Reduce the music relative to the voice as a separate editing change, then measure and listen again under the same conditions. If reducing overall level makes speech disappear, stop treating master volume as the only control. Resolve the balance and intelligibility before declaring the soundtrack ready.

Approve the published version separately

A publisher may process uploaded audio, so review the actual posted preview after the production file has passed its checks. That preview is a different review stage, not evidence that every viewer will hear identical volume. Retain the delivery file and record any observed difference without inventing a platform normalization rule.

CartVisual’s current assembled TVC is silent. Voice recording, music balance, and loudness measurement are external tasks, not built-in site controls. Stop delivery when the required target or audio balance remains unresolved. A filename such as broadcast-safe is not proof of a measured compliant file.

Audio review record

Use this matrix as a manual review record and add your own real evidence. Its conditions are not completed test results.

Download this question’s review worksheet (CSV)