A useful speech comparison can become inconclusive when one candidate plays louder, through different speakers or from a different review copy. Before commissioning another revision, make the comparison repeatable. The aim is to let reviewers describe the sound treatment they prefer rather than an accidental change in the way they heard it.
Adobe's Adobe filter and equalizer reference includes output gain compensation for level changes introduced by tonal adjustment. This is a practical reminder that processing and overall playback level are separate variables. The following review method is commissioning guidance, not a calibration standard or a prescribed listening level.
Establish one comparison set
Ask the editor to prepare the same excerpt in the original and proposed treatments, with levels made appropriately comparable. Use unambiguous version labels and retain the exact review links or files. Reviewers should not have to reconstruct which candidate they heard from a message saying the latest one sounds better.
Choose an excerpt that contains the concern under discussion and enough normal speech to judge context. If the decision is about a particular repair, include that repair. If it is about the character of a whole voice, a tiny fragment will provide too little basis for approval.
Keep the first comparison simple
Listen to the candidates on the same device without changing its playback setting between them. If a change is needed for comfortable review, restart the comparison consistently rather than applying the new setting to only one version. No numerical listening target is required for this buyer-level process.
Ask reviewers to state the passage, the candidate and what changed. The background is less intrusive but the last word sounds metallic is an actionable observation. Candidate B is better offers a preference but gives the editor little information about why it should guide the remaining work.
A hypothetical divided review
Suppose two stakeholders disagree about a cleaned interview. One reports that the revision sounds clearer; the other says the voice has become thin. It emerges that the first heard an older, louder review file on a desktop system while the second heard the new candidate on a phone.
The producer supplies a controlled comparison set. Both reviewers now hear the same candidates on their respective devices and identify the same word as the source of concern. The editor can address that local issue instead of reversing a whole treatment in response to apparently contradictory preferences. This is a fictional example, not a Live Images review record.
Use additional devices to answer a different question
After choosing a candidate under consistent conditions, assess whether it remains acceptable through the intended audience's ordinary playback routes. Keep this separate from the initial treatment comparison. Otherwise every device change can restart the question of which processing choice was better.
Record a device-specific concern as such. If a word is unclear only on one route, the producer can investigate that observation without treating it as proof that the entire audio treatment failed. Different listening contexts may expose different limitations; the brief should identify which contexts matter to the film.
Acceptance questions include whether reviewers used the same source version, whether candidate levels were made comparable, whether the preferred result survives the agreed audience check and whether remaining objections name a repeatable moment. Establish who consolidates the decision. A small amount of review discipline can turn vague requests for more polish into a clear instruction that the editor can implement and verify.
Discuss the sound brief
Contact Live Images with the candidate versions and conflicting review notes to discuss a practical, bounded sound review process.
Prepared with AI assistance for Live Images. Planning guidance is general; suitability and production scope must be agreed for your project.