How to Edit an Interview Faster: The Transcript-First Workflow

The fastest way to edit an interview is to stop editing on the timeline and start editing on the transcript: read the words, highlight the strongest lines, arrange them on the page, then push that selection straight to your timeline. The bottleneck in interview editing was never the dragging of clips. It was the scrubbing, the re-watching, the trying to remember which answer was the good one. Reading fixes that.

Here is the uncomfortable math. People read silent text at roughly 238 words per minute for non-fiction, according to a 190-study meta-analysis by Marc Brysbaert at Ghent University, while conversational speech lands closer to 150 wpm. So scanning a transcript is well over a third faster than listening back at real time, and you can skim, jump, and re-read in a way you simply cannot do with a scrubber. That gap is the whole reason this workflow exists.

Why scrubbing the timeline is the slow part

Open a 90-minute interview in your NLE and the cost is hidden in plain sight. To find the one clean version of an answer, you play it. To compare two takes, you play both. To remember whether the subject got to the point on the third pass, you play it a third time. None of that is editing. It is searching, at one times speed, with your hand on the spacebar.

A transcript turns that search into a read. You see all six attempts at the same answer stacked as text, pick the one that lands, and move on. The footage has not changed. What changed is that you are deciding with your eyes instead of your ears, and your eyes are faster.

The transcript-first workflow, step by step

1. Transcribe with word-level timecode

This is the part that makes everything else work. A plain transcript is a reading document. A transcript with timecode on every word is an edit. When each word is anchored to a frame, highlighting a sentence is the same as marking an in and an out point. Skip this and you are back to manual logging.

2. Read once, all the way through

Resist the urge to start cutting on line one. Read the full transcript first, the way you would read a script before blocking a scene. You are looking for the spine: the two or three things this person actually said that the piece is about. Everything else is in service of those.

3. Highlight your selects

Now go back and mark the strongest lines. These are your selects. Be ruthless and be generous at the same time: ruthless about cutting rambling and throat-clearing, generous about keeping a line you are unsure of, because deselecting later is trivial. In ScriptCut you highlight right on the transcript and each highlight becomes a precise cut, since the word-level timecodes are already attached.

4. Trim the fillers and false starts

Within a kept line, drop the ums, the repeated words, the abandoned sentence the subject restarted. This is where a line goes from raw to broadcast. Do it on the words and the video follows.

5. Arrange the story on the page

Reorder your selects into the sequence that tells the story, not the order they were recorded in. A great answer from minute 70 might be your opening line. You are building the narrative as a list of text blocks before a single clip touches the timeline.

6. Verify tone on the clip, then export

This is the step skeptics are right about, so do not skip it. Reading tells you what was said, not how. Play each selected line to confirm the delivery matches what you read, then export the arrangement as a timeline to DaVinci Resolve or Premiere Pro, Final Cut, or Avid, and finish the craft work there.

A worked example

Say you have a 60-minute founder interview to cut to a 4-minute piece. The old way: you log it, watching the whole thing with a notepad, which is 60 minutes minimum, then you assemble selects on the timeline, scrubbing back and forth, call it another 90 minutes of searching before you have a rough cut. The transcript way: you read the 60 minutes of talk in about 25 minutes, highlight selects as you go, reorder the blocks in 15 minutes, then spend your real attention on the part that matters, the verify-and-finish pass. You did not work less hard. You spent the hard work on craft instead of search.

Common mistakes

Editing the timeline and the transcript in parallel. Pick one source of truth. If your selection lives on the page, make every structural decision there, then go to the NLE once to finish. Bouncing between the two is how you lose track of what is in.

Trusting the read for tone. A line can read perfectly and land flat, or a sarcastic aside can look hostile in plain text. Always play your selects.

Over-tightening. Cut every breath and the subject sounds like a machine gun. Leave the natural beats. Clean is the goal, not airless.

Stitching words the subject never said in sequence. Reordering whole answers and tightening within a line is normal editorial work. Splicing fragments from different moments into a sentence they never spoke is the Frankenbite, and it is an ethics line, not a style choice. More on that below.

The honest tradeoffs

Errol Morris, who has shot some of the most rigorous interviews in documentary, is blunt about this approach. In a 2003 Transom interview he said, 'No, I don't edit from the transcripts, ever. I edit from the film.' He is not wrong that a transcript flattens performance. The fix is not to abandon the speed of reading, it is to treat your transcript selects as a hypothesis and confirm each one on the clip before you commit. That is why word-level timecode and the ability to play any line matter so much: the page proposes, the footage decides.

The other tradeoff is setup. You need a clean transcript with real timecode, which costs a few minutes up front. On a five-minute clip that overhead barely pays off. On anything over twenty minutes it pays for itself many times over.

The takeaway

Interview editing feels slow because we have been doing the deciding at playback speed. Move the deciding to the transcript, keep the craft on the timeline, and verify tone in between. You will not just finish faster. You will make better structural choices, because for the first time you can see the whole interview at once. When you are ready, do the next pass even faster with a tighter soundbite process and a clean way to organize your footage.

Sources

• Brysbaert, How many words do we read per minute? A meta-analysis (2019)
• Errol Morris on his editing process, Transom
• Adobe, Text-Based Editing in Premiere Pro (NAB 2023)
• No Film School, documentary editing resources