How to edit an interview video: structure, cutaways and sound

How do you edit an interview video?

Work from the words to the picture, in this order:

Transcribe everything. You cannot hold two hours of answers in your head; you can search and highlight a transcript.
Find the story. Mark the answers that say something only this person can say, and arrange them into a beginning, a middle and an end.
Cut the questions. Let the subject's answers carry the piece, unless the question itself is part of the story.
Tighten each answer. Take out false starts, fillers and digressions, keeping the pauses that carry thought.
Cover the cuts. Every cut inside an answer is a jump; hide it with B-roll, a second angle or a punch-in.
Fix the sound. Clean each voice, level them to one loudness and lay room tone under every gap.
Add names and captions. A lower third at each speaker's first appearance, and captions for viewers with the sound off.

How do you find the story in interview footage?

In the transcript, not the timeline. Editors call this a paper edit: highlight the strongest answers, write them out in a new order, and only then cut picture.

Look for:

The line only they could say. A detail, a memory or an opinion, not a summary anyone could give.
Answers that stand alone. "It was 1987 and the plant had just closed" beats "Like I said, it was then".
Turns. A change of mind, a surprise, a moment where the voice changes.
An ending. The answer that resolves what the opening asked.

Sharpy transcribes on your Mac with two speech engines that vote, and a word the engines were unsure of is underlined in the words view so you check it before it ends up in a caption. ⌘F searches what is said in every recording, ignoring case and accents, so "where she talks about the factory" is one search. Sharpy can also build a sound-only radio cut from the transcript first, across several recordings, so the argument holds before any picture goes on.

How do you cut out the interviewer's questions?

Split the recording by speaker, then delete the interviewer's turns. Sharpy finds who speaks when, so you can cut every question from the interviewer; tell it how many people speak and that count is exact. In the words view, select the words and press Delete, and they leave picture and sound together; right-click puts them back from the original media.

Two things make this work:

Ask for full-sentence answers when you shoot. "What did you do next?" answered with "Then I sold the farm" survives without the question; "Sold it" does not.
Keep a question when the answer needs it. If the answer only makes sense after it, keep the question as a caption card or as audio over the subject listening.

Then tighten each answer. Sharpy's review list finds fillers, repeated words, restarts, retakes and long pauses, each with what an editor would do and why, and it shortens a pause to the speaker's own rhythm rather than closing it. Removing filler words and dead air covers those rules.

How do you use cutaways and B-roll in an interview?

Put them over cuts, and over anything the subject describes. A cutaway lets you cut inside an answer without the jump showing, and B-roll shows what words alone cannot.

In a Sharpy cut script, declare a source at the top, [source a = /path/to/broll.mov], and on a line write show a 4.0 to cut to it from 4.0 seconds in. The voice, captions and on-screen text carry on untouched; only the picture changes, and the cutaway runs on across the line instead of restarting. When you have no footage of your own, Sharpy searches Pexels, Pixabay, Wikimedia Commons, NASA's library and museum open-access collections in one list, fetches each clip at the smallest size at least as large as your piece, and writes the credits it owes into a file for the video's description, never onto the picture. Then match the clip's colour to the interview: Sharpy measures both and reports how much of each gap it closed. What is B-roll covers what to shoot and where it goes.

Reactions work like cutaways. A nod or a laugh from the interviewer, held over the subject's words, is an L cut; J cut and L cut explains both.

How do you edit an interview shot on two cameras?

Sync the cameras, cut the conversation on one sound track, then choose the angle for each stretch: the close-up for the answer, the wide for the room, the interviewer for a reaction. Sharpy lines a second angle up by what both microphones heard and takes out the drift between their clocks; how to sync audio and video covers that step. It then turns each speaker's turns into angle cuts, or crops around each speaker from a wide shot, adds a lower third at each speaker's first turn, and you can switch any angle after the cut. Multicam editing goes through it in full.

With one camera, punch in: a 15 to 25% change in frame size across a cut reads as a second camera.

How do you make two voices sound like one programme?

Interview sound fails in three places: one voice louder than the other, a noisy room, and silence where a cut took the room away.

Problem	The fix	In Sharpy
One voice quieter	Level each clip's speech to one target	Brings each clip's speech to −23 LUFS, riding gently within the clip so the room in pauses is not lifted
Room noise	Clean the voice without losing words	Tries cleaning strengths and keeps only one that keeps 97% of the words
Dead silence at cuts	Fill every gap with room tone	Lifts the room from the recording's quietest stretches and lays it on its own track

On two lines recorded 17.5 dB apart, at −34.6 and −17.1 LUFS, Sharpy's dialogue levelling brought each to −23.0, in 93 milliseconds for both six-second clips on the base Mac mini M4. It gives at most 12 dB of gain either way; anything further is reported as a recording to fix. Ask the agent with adjust_clip and level set to "dialogue".

A man adjusts a microphone on a boom stand in a dim, warm room hung with chandeliers, a woman with a violin waiting at one side. — Thirty seconds of the room with nobody speaking gives the editor something to fill every gap with.Photo by Kyle Loftus on Pexels

Room tone is the quiet sound of the place itself. Record thirty seconds of it on the day, with everyone still. When a cut leaves a hole on the voice track, Sharpy fills it with the room, lifted from the recording's quietest stretches, never reusing a stretch within 20 seconds; where those run out, it makes the room again from its own hum and noise, and a minute of it never repeats. In the window, Clear, Leaving the Gap takes a stretch out and leaves the hole to fill.

Frequently asked questions

How long should an interview video be?

As long as its best material. A ten-minute answer reel from a two-hour interview is common; a short for social is one answer that stands alone.

Should I show the interviewer?

Only when the exchange is the story, as in a podcast or a confrontation. In a documentary-style piece, cutting the questions keeps the focus on the subject.

What is the best way to film an interview for editing?

Ask for full-sentence answers, record room tone, use a second camera or shoot in 4K for punch-ins, and capture plenty of B-roll of what the subject talks about.

How do you edit out ums from an interview?

Cut the ones that break a sentence and keep the pauses that carry thought. Sharpy's review list finds them and lets you hear, cut or keep each one.

Is my interview footage uploaded anywhere?

No. Transcription, speaker splitting, cutting, levelling and rendering all happen on your Mac.

Sources

Checked on 18 October 2026. Stock sources Sharpy searches, each credited in the video's description where its licence asks: