How to transcribe an interview

To transcribe an interview, you turn the recording into a written document that shows who said what. The process is the same whether you’re a journalist, a researcher or a hiring manager: record well, choose a transcription style, create a first draft, then review it against the audio. This guide walks through each step and ends with an interview transcription example you can copy.

Step 1: Record the interview properly

A good transcript starts with a good recording. Most transcription problems — missing words, guessed names, “[inaudible]” every few lines — come from the audio, not the transcriber.

• Put the microphone close to both speakers. A phone on the table between you works; a laptop across the room doesn’t.
• Pick a quiet room. Cafés, air conditioning and open windows all compete with voices.
• Record a test. Ten seconds of “testing, testing” saves an hour of guessing later.
• For calls, record both sides. Zoom, Teams and Google Meet can record the meeting; save the audio file afterwards.
• Ask for spellings. At the end, ask the interviewee to spell their name, company and any unusual terms.

Step 2: Choose a transcription style

Before you type a word, decide how literal the transcript should be.

Style	What it keeps	Best for
Full verbatim	Every word, “um”, “uh”, false start, repetition, laugh and pause	Legal work, linguistics, psychology research
Clean verbatim	Every meaningful word; fillers and stumbles removed	Journalism, qualitative research, business interviews
Edited	The meaning, rewritten into tidy sentences	Published Q&As, blog posts, summaries

Most people want clean verbatim. It reads naturally but stays faithful to what was said, so you can still quote it.

Step 3: Create a first draft

There are two ways to get the words on the page.

Typing it yourself. Play a few seconds, pause, type, repeat. A foot pedal or keyboard shortcuts for pause and rewind help. Expect four to six hours of work per hour of audio.

Using AI transcription. Speech recognition now gets clear recordings mostly right, which turns transcription into an editing job. With Corbeia you can transcribe the interview audio to text for free in your browser: drop in the recording, wait a few minutes, and download the transcript. iPhone recordings are usually M4A files — see M4A to text — and video calls can go through MP4 to text.

Because Corbeia transcribes on your own computer, the recording is never uploaded. That matters when interviewees were promised confidentiality or your ethics approval restricts where data can go.

Step 4: Review and edit against the audio

An AI draft is a draft. Play the recording at normal or slightly faster speed and read along.

1. Fix names and jargon first. Speech recognition guesses at unusual words; use the spellings you collected.
2. Add speaker labels. Mark each change of speaker, for example Interviewer: and Participant:.
3. Mark what you can’t hear. Use [inaudible 12:34] with a timestamp so you, or someone else, can check it later. Never guess.
4. Remove fillers if you chose clean verbatim.
5. Add timestamps every few minutes or at each new question, so quotes are easy to find.

Step 5: Format the transcript

A consistent layout makes a transcript easy to read and cite.

• A header with the interview title, date, location or platform, participants and duration.
• Speaker labels in bold or capitals, followed by a colon.
• A new paragraph for every change of speaker.
• Square brackets for anything that isn’t speech: [laughs], [crosstalk], [pause].
• Anonymized names where required: Participant 3, [Company name].

Interview transcription example

Here is an example of transcribing an interview in clean verbatim, with speaker labels and timestamps:

Interview: Remote work and team habits
Date: 14 October 2026 · Zoom · 32 minutes
Participants: Interviewer (I), Participant 4 (P4)

[00:00:12]
I: Thanks for joining. Could you start by describing a typical workday?

P4: Sure. I start around eight, and the first hour is just messages —
catching up with the people in Singapore before they log off.

I: Is that hour useful, or is it mostly noise?

P4: [laughs] Honestly, half and half. The useful part is the decisions.
The rest could be an email.

[00:04:30]
I: You mentioned decisions. How does your team make them?

P4: Mostly in a shared doc. Someone writes the proposal, and everyone
has 24 hours to comment. If nobody objects, it goes ahead. [inaudible 00:04:51]
works well for small things, less for big ones.

The same passage in full verbatim would keep “um, so, I, I start around, uh, eight” — accurate, but much harder to read.

Tips that save hours

• Transcribe soon after the interview, while you still remember unclear passages.
• Keep a glossary of names and terms for multi-interview projects.
• Work in passes: first fix words, then speakers, then formatting.
• Keep the original audio until the project is finished — the transcript is only as trustworthy as your ability to check it.
• Use the transcript with AI carefully. Pasting it into a chatbot for a summary is useful, but check quotes against the transcript, not the summary.

Summary

Record close and quiet, pick clean verbatim unless you need every “um”, get a first draft with free audio to text transcription, then review it against the recording, add speaker labels and timestamps, and format it consistently.