How to Transcribe Interviews for Your Dissertation or Thesis (2026 Guide)

Short answer: record each interview as a separate file, convert it to text with an automatic transcription tool, review it while listening to the audio, and anonymize personal data before including it in your work. That way a one-hour interview takes you under an hour in total, instead of the 4–6 hours it takes to transcribe by hand.

Here's the full process, from before you record to how to cite transcripts in your dissertation.

1. Before recording: consent and interview guide

• Informed consent. Ask for written permission to record and to transcribe with digital tools. If your department has an ethics committee, use its template.
• Participant codes. Give each person a code (P01, P02…) from the start and use it in file names, so personal data never sits next to your recordings.
• Semi-structured guide. Having your questions in front of you helps you avoid interrupting, and overlapping speech is what transcribes worst.

2. Record so the transcript comes out well

Transcript quality depends mostly on the recording:

1. Place your phone or recorder less than a meter (3 ft) from both people.
2. Find a small, quiet room. Cafés and echoey spaces multiply errors.
3. For video-call interviews, record the call audio (Zoom, Teams or Meet) instead of recording the speaker with another device.
4. Do a 30-second test and listen to it before you start.

If your interviews are already recorded and not perfect, that's fine: they still transcribe, you'll just review a little more.

3. Transcribe with AI

1. Open the transcription tool and upload the file (MP3, M4A, WAV, MP4…).
2. Choose the interview language.
3. Press Transcribe. Within minutes you'll have the full text with timestamps.

If you haven't tried it yet, you get 15 free minutes with no sign-up, enough to check the quality with a real excerpt of your interview. For a master's dissertation with 8–10 interviews, people usually buy a minute pack: see pricing, no subscription and no expiry.

4. Review while listening

Automatic transcription is very accurate with good audio, but not perfect. Check especially:

• Proper names, acronyms and numbers.
• Technical terms from your field.
• Passages with overlapping voices.

In the editor, clicking a sentence plays that clip, so corrections are quick. The tool doesn't label speakers automatically yet: add "Interviewer:" and "P01:" while reviewing (timestamps make each turn change easy to find).

5. Decide on the level of verbatim

Check with your supervisor which type of transcript your methodology requires:

Type	What it includes	When it's used
Verbatim	Everything: repetitions, fillers, pauses	Discourse or conversation analysis
Intelligent (clean) verbatim	No fillers or hesitations, but the way people speak is kept	Most thematic analyses
Summary	Only the relevant content	Almost never acceptable as data

Automatic transcription is closest to clean verbatim. If you need strict verbatim, add relevant fillers and pauses while reviewing.

6. Anonymize

Before sharing the transcript with your supervisor or adding it as an appendix, replace names, places, companies and anything that could identify the person: "[name]", "[city]", "[energy company]". Do it in the editor before downloading, or in Word afterwards.

Uploaded files are deleted automatically after 7 days and never used to train models; still, keep the originals wherever your university requires.

7. Download and import

Download the transcript as Word and name it with the participant code (for example, P01_2026-03-12.docx). NVivo, ATLAS.ti and MAXQDA import Word documents directly.

8. How to cite interviews in your dissertation

• In the methodology, explain how you recorded, that transcription was done with automatic speech recognition and that you reviewed it manually.
• In the text, quote excerpts with the participant code: "…we had never looked at it that way" (P03).
• In the appendices, include anonymized transcripts if your department asks for them.

FAQ

Is it acceptable to use AI to transcribe dissertation interviews? Yes, as long as you review the result and explain it in your methodology. It's increasingly common; check your department's specific rules.

How much does it cost to transcribe ten one-hour interviews? That's ten hours of audio. With pay-as-you-go you only buy those minutes, no subscription; see the packs on the pricing page.

What if an interview is in another language? Choose that language when uploading: more than 60 are supported. If it mixes two languages, pick the dominant one.

More on the process in the research transcription and interview transcription guides.