Talking Head & Interview Editing
ChatCut can shorten pauses and remove unwanted speech ranges from supplied talking-head footage. Choose the dialogue track, scope an Agent request or use the Transcript controls, and keep the result editable. Listen through every join before export: shorter silence is not automatically better pacing, and filler words can carry meaning. Restore cuts that clip syllables, breaths or important context.
Speech-editing controls, last verified 2026-10-28
The table records documented controls checked against beta source and public guides. It is not a tested before-and-after example or a guarantee that every spoken word will survive an automatic pass.
| Operation | Documented behavior and review needed | Source |
|---|---|---|
| Select dialogue | Choose the intended transcript track; deleting selected words cuts their time range. | Transcript and Captions |
| Adjust pauses | The Pauses control applies a length across the selected transcript track. It can restore existing source pause time, not invent silence. | Transcript controls |
| Request cleanup | Specify the target passage, what may be removed, what must remain and the desired pacing. | Talking Head Edit |
| Isolate speech | Right-click selected spoken-video or audio clips → AI Voice Isolation; Restore original audio reverses that treatment. Listen before keeping it. | Voice isolation |
| Review and deliver | Inspect the editable timeline, use Undo or Versions for revisions, then open the exported file. | Review workflow |
Shorten silence with a defined scope
Import the recording, wait for processing and open Transcript. Use Workspace → Transcript if needed. Choose the track with the intended speaker, especially when several camera or audio tracks contain dialogue. For a passage-specific request, attach its transcript text or clip with Selection Mode.
Clean up only this selected passage. Remove clear false starts and repeated takes. Shorten pauses longer than 0.8 seconds while keeping natural breathing, the complete argument and its caveat. Use only the supplied recording. Put the editable cut on the timeline for review; leave captions and export for later.The 0.8-second value is an example instruction, not a universal quality setting or a measured default. Pick the threshold for the speaker and material. A reflective interview needs different breathing room from a concise tutorial.
For manual control, select an unwanted spoken range and press Backspace or Delete. Use Pauses for a track-wide pause-length pass; review its full scope before applying. To target only one moment, use a selection-based edit instead. Clip view lets you reorder the corresponding clips.
Decide what counts as filler
Words such as “like” and “well” can affect meaning. Give the Agent examples of removable hesitations and name any qualifications, numbers or emotional pauses that must stay. Start with clear false starts and duplicated takes, then decide whether smaller hesitations need cutting.
Do not delete Transcript words simply to fix spelling in captions. That action cuts speech. Correct caption wording in the active source’s Content panel; see AI Captions.
Listen before accepting the first cut
Play the whole selected passage with sound on, including both sides of each cut. Compare it with the original recording where a join is uncertain. Check complete syllables, natural breaths, unchanged meaning and picture continuity; a duration change or completion message cannot establish these qualities.
Use Undo for an unwanted edit, or save a Version before a larger pass. A precise revision is easier to judge: “Restore the breath before the final sentence; keep the other cuts.” Keep the original media and the editable project.
Review voice isolation separately
If background noise distracts from speech, select the relevant clips, right-click and choose AI Voice Isolation. Wait for Isolating voice… to finish and listen for intelligibility, unnatural speech and intentional room sound that may have been removed. Use Restore original audio when the treatment does not help.
This treats spoken audio, not picture noise. Keep music and ambience separate. If a later trim extends beyond the processed range, follow the warning and reapply isolation to cover the new range. Processing completion alone does not prove improved sound.
Finish and export
Once the speech works, add and review captions, then check the target aspect ratio and any B-roll. Export the active Timeline with Export → Video. Open the delivered file and listen again before sharing. Use Export → Subtitles separately for SRT or TXT. Agent editing can consume credits; see Credits.
Removing silence automatically
Frequently asked questions
How to remove silence from a video automatically?
Sources and verification scope
Last verified 2026-10-28 against Talking Head Edit (updated 2026-10-24), Transcript and Captions (updated 2026-10-10), Voice Isolation (updated 2026-10-19) and the export guide (updated 2026-10-24). No auditioned silence-removal or isolation before-and-after result, speed improvement or word-preservation guarantee is claimed.
