Skip to main content

How to Review Dialogue Cleanup in a Client Session

· 5 min read
Creative workflow guides from FlowLabs

The cleaned interview sounds clearer until the speaker finishes a sentence. The room sound drops out under the last word, then returns on the next cut. The client likes the voice but hears a join they could miss if you played the clip by itself.

For a dialogue cleanup review, cue the original and treated versions of the same passage, match the perceived level of the voice, and play both with the surrounding edit. Ask whether the line is easier to follow and whether anything else changed enough to pull attention away from it.

Choose a passage that includes the problem​

Start with the actual listening problem. A quiet line buried by a fridge hum calls for a different test from an interview recorded in a reflective room. Write down what you want the listener to be able to hear. “The last four words should be intelligible under the score” gives the client a specific decision.

Pick a short passage with the difficult words, a little space before and after the line, and a cut into the next shot. The silence between phrases can expose changes in the room sound that are easy to miss while the speaker is talking. Include the music or effects that will play in the delivered edit.

Adobe's November Premiere update gives editors controls for dialogue, music, ambience, and sound effects separated from a clip. In Premiere 27.0, the controls are under Properties → Audio → Enhance Audio. If you used those controls to treat an interview, save the settings for the version you present. The comparison below also works with a manual cleanup or a separate dialogue processor. Adobe's Enhance Audio guide · Premiere 27.0 release notes

Match the voice level before playing the pair​

Set up two versions of the passage with the same picture, music, and other tracks. Keep them at the same position in the timeline. If the treatment is on an alternate dialogue track, check that only one dialogue track is audible at a time.

Listen to a representative phrase and adjust the output gain of the treated dialogue until the voice feels close in level to the original. A meter can flag a large difference, and a short listening pass helps you judge how loud the voice feels over the music. Leave playback volume alone when you switch versions. If the scene needs a louder voice in the final mix, audition that as a separate change after you choose the treatment.

Start both plays a little before the line. Let each version run through the next cut before you ask for feedback. When the client wants another pass, you can reverse the playback order and keep the same start point.

Listen to the words and the cut​

  • Speech. Can you follow the line on the first listen? Pay attention to quiet consonants, breaths, and word endings. If a word seems clipped, replay the full phrase and see whether the same word disappears again.
  • Room sound. Does the space still feel continuous when the speaker pauses? Listen to the gap before the first word and the tail after the last one. Compare those moments against the shots on either side.
  • Music and effects. If they were already mixed into the clip, check whether a change intended for the voice also affects music or an audible action. Listen for the change in the full scene, where it can affect the timing or feel of a cut.

Play at normal speed for the decision. If a fine detail is hard to judge in the review, mark the timecode and check the source and treated audio on your finishing setup. Share a reference export for a separate listen if the decision depends on that detail. When to use live and async review

Write down the treatment the client chose​

Suppose an interviewee says “we opened in May” under a music cue. The first pass reduces fan noise and makes “May” clearer, but the room sound fades as the speaker pauses. You ease off the treatment and keep the existing music level. The client hears the line clearly in the second pass and the cut still feels continuous. Save the second pass as its own version so the mixer can recreate the approved choice.

Use this note in the review log and replace the example values with your own.

Sequence and range — Interview cut v07, 01:04:12:00–01:04:24:00

Versions heard — Original and dialogue treatment v02, voice levels matched, score in place

Decision — Keep v02 for this line. “May” is clear and the room stays audible through the cut.

Follow-up — Check the adjoining interview line before applying v02 settings to it.

If the client hears the voice twice even when you play the original, check the review audio paths before recording a mix decision. How to isolate echo in a remote review

For your next client session, send timeline playback into Flow from a WHIP encoder and use the room microphone to discuss each pass after playback. Flow encoder setup · Run a live review in Flow