Skip to content
Guide

How to Transcribe an Interview (and Make Sure It's High Quality)

Cover image with the title "How to Transcribe an Interview (and Make Sure It's High Quality)" on a lavender-to-blue gradient with the Oryn app icon
Guide cover image for this Oryn blog post, showing the article title on the Oryn lavender gradient.

A good interview transcript starts before you press record. Here's the full process, from consent and microphone placement to checking every quote.

By 4 min read
In this article

TL;DR

To transcribe an interview with high quality, get consent, pick a transcript style before you start, record close to each speaker in a quiet room, run the audio through an AI tool that labels speakers, then check every name, number and quote against the recording. Clean audio and a careful review matter more than which tool you use.

Ask permission on the recording itself. Start with "Do I have your permission to record this conversation?" and keep their answer at the top of the file. That gives you proof of consent attached to the interview.

The law sets a minimum. US federal law allows recording when at least one person in the conversation consents, but about 11 states require everyone's consent. Some states split the rule: Missouri and Oregon require all parties' consent for in-person conversations but only one for phone calls, while Connecticut and Nevada are the reverse. If you and the interviewee are in different states, follow the stricter rule.

Research, journalism and HR interviews often add their own rules on top. A university study may need a signed consent form approved by an ethics board, and a newsroom may require on-the-record agreement. Check these before booking the interview, not after.

Step 2: Choose your transcript style

Decide how much of the speech to keep before the first interview. Switching styles halfway through a project means redoing the earlier transcripts so they match.

Style What it keeps Use it for
Strict verbatim Every word, "um", repeat, false start and pause Conversation analysis, legal records
Clean verbatim Every real word, with fillers and stutters removed Most research interviews, journalism, user research
Edited The meaning, with grammar and flow tidied Published Q&A articles, reports

Clean verbatim is the practical default for most interviews. It reads easily and stays faithful to what was said. Most AI transcription tools produce something close to clean verbatim, so plan extra work if you need strict verbatim.

Write your choice down in a one-line style guide, with how you'll mark unclear audio and speaker names, so every transcript in the project follows it.

Step 3: Record clean audio

The recording sets the ceiling on transcript quality. No software recovers words that the microphone didn't capture.

For an in-person interview:

  • Pick a small room with soft surfaces. Carpets and curtains cut echo; glass and bare walls add it.
  • Place the microphone at equal distance from both speakers, or slightly closer to the interviewee, since their answers matter more than your questions.
  • Use a clip-on microphone in echoey rooms. It picks up less of the room than a phone on the table.
  • Record 30 to 60 seconds, then play it back before you start.

For a remote interview:

  • Both of you should use a headset or external microphone, kept about 4 to 8 inches from the mouth.
  • Turn on separate audio tracks per speaker if your call software offers it.
  • Run a backup recording on a second device in case the main one fails.

Step 4: Transcribe with AI, a human, or both

AI gets you a draft in minutes; a person gets the hard parts right. Typing a transcript by hand takes several hours per hour of audio. University guides commonly put it at four to six hours, depending on audio quality.

For most interviews, a hybrid works best: an AI tool produces the draft, and you correct it in Step 5. Look for a tool that:

  • Labels speakers automatically, so you aren't assigning every line by hand.
  • Has no length cap that cuts off a long interview.
  • Supports the interview's language.
  • Exports to a format you can edit, such as TXT or PDF.

Oryn covers these on iPhone, iPad and Mac. It labels each speaker, has no limit on transcription length, writes a summary of the key points, and exports to PDF or TXT. For legal records or broadcast quotes where every word matters, a professional human transcriber is still the safer choice.

Step 5: Review and correct the transcript

Never publish or analyze an AI transcript you haven't checked against the audio. The errors that matter most are the ones that look right.

  1. Listen through once at normal speed with the transcript open, fixing as you go.
  2. Check every name, number, date and technical term. These are the most common mistakes.
  3. Confirm speaker labels, especially where people interrupt each other.
  4. Mark what you can't hear with a tag and timestamp, such as [inaudible 00:12:43], instead of guessing.
  5. Mark overlaps with [crosstalk] where two people speak at once.
  6. Replay every quote you plan to use and confirm it word for word.

If a passage is unclear and the quote matters, contact the interviewee and confirm what they said. A short follow-up email is better than a misquote.

Step 6: Store and share it safely

An interview transcript is often more sensitive than it looks. It can hold names, workplaces, health details or opinions the person wouldn't want linked to them.

  • Remove identifiers for research. Replace names, places and employers with codes or pseudonyms, and keep the key in a separate, protected file.
  • Store recordings and transcripts in one access-controlled place, not scattered across email and downloads.
  • Share the transcript, not the audio, when someone only needs the content.
  • Delete the recording when you no longer need it, following your consent form or your organization's retention rules.

Name files consistently, for example 2026-11-09_interview_P04_clean.txt, so anyone on the project can find the right version. For a full secure workflow, including extra steps for journalists, see How to Transcribe Sensitive Interviews and Conversations Securely.

References

Frequently asked questions

Does Oryn label who is speaking in an interview?

Yes. Oryn adds speaker labels to every transcript, so the interviewer's questions and the interviewee's answers are separated automatically. Check the labels where people interrupted each other before you quote anyone.

Can Oryn transcribe a long interview?

Yes. Oryn has unlimited transcription, so a two hour interview or a full day of back-to-back sessions transcribes without hitting a cap. You don't have to split the recording to stay under a limit.

Can Oryn transcribe a remote interview on Zoom or Teams?

Yes, on a Mac. Oryn records the computer's own audio, so it captures both sides of the call without joining as a bot. Ask for consent at the start of the call, since the other person gets no automatic notice.

How do I export an interview transcript from Oryn?

Export it as a PDF or TXT file. PDF suits sharing a fixed copy with an editor or client. TXT suits pasting quotes into an article or importing the text into research software.

Keep reading