If you are conducting qualitative interviews for a thesis, dissertation or research project, transcription is unavoidable. The recordings contain your participants' voices and often personal stories they shared in confidence. That makes transcription a privacy question as much as a time question. This guide walks you through how to transcribe research interviews securely and in line with what review boards typically expect.
Why interview recordings need special care
A voice recording is identifiable information. People can be recognized by their voice, and participants often mention names, employers or places. Interviews may also touch on sensitive topics such as health, religion or immigration status. In the United States, research with identifiable private information falls under your Institutional Review Board (IRB) and the Common Rule. If you work with participants in the European Union, the GDPR applies as well and treats such recordings as personal data.
Either way, the principle is the same: collect only what you need, protect it while you have it and delete it when you are done.
What to include in your consent form
Most universities provide a consent template, so check with your IRB office first. A strong consent form covers:
- that the interview will be recorded and transcribed
- who will have access to the recording, including any outside transcription service
- where recordings and transcripts will be stored
- how long they will be kept and when they will be destroyed
- that names and identifying details will be removed from the transcript
- that participation is voluntary and participants can withdraw at any time
The second point matters most. If your consent form says only the research team will hear the recording, uploading it to an online service afterwards would break that promise.
Three ways to get a transcript
| Transcribe by hand | Online service | Local software | |
|---|---|---|---|
| Where the recording goes | stays on your device | uploaded to the provider's servers | stays on your device |
| Time per hour of audio | often 4 to 8 hours | a few minutes plus editing | a few minutes to an hour plus editing |
| What you need for compliance | secure storage | IRB approval of the vendor, often a confidentiality or data processing agreement and disclosure in the consent form | secure storage |
Online services are fast, but the provider processes and stores your recordings. Many IRBs require that vendors be approved in advance or sign a confidentiality agreement, and studies involving health information may bring HIPAA into play. Under the GDPR, you need a data processing agreement, plus additional safeguards if the servers are outside the EU.
Local software avoids these issues entirely: speech recognition runs on your own computer, so the recording never leaves it. Stillscript works this way and runs directly in your browser with nothing to install. You can verify it yourself: after the first load, transcription keeps working with your Wi-Fi turned off.
Step by step
- Check your protocol: confirm what your IRB approved for recording, transcription and storage.
- Get consent: before the interview, in writing or by email for remote interviews, and file it securely.
- Record: on your phone or laptop with cloud sync turned off. Check whether your voice memo app backs up recordings automatically.
- Transcribe: by hand, locally with software or with a service your institution has approved.
- Review: speech recognition makes mistakes, especially with technical terms, names and accents. Listen again where it matters.
- De-identify: replace names, places and other identifying details in the transcript (see below).
- Delete: destroy the recording once you no longer need it, no later than the date in your consent form.
De-identifying your transcript
Nobody reading your appendix or dataset should be able to identify your participants. Replace names with codes and make details more general:
Original"I've worked at Miller's Bakery in Portland since 2019, and my manager Sarah Johnson …"
De-identified"I've worked at [bakery] in [city in the Northwest] since [year], and my manager [Person A] …"
In Stillscript, use “Replace names” above the transcript to swap a name for a code everywhere at once. Keep the key that links codes to real names separate from the transcripts and protect it like the recordings themselves. Your field or advisor may also expect a specific transcription style, for example clean verbatim or full verbatim with pauses and filler words. Ask before you start.
Storing recordings securely
- Store recordings and transcripts on your own device or in storage your institution has approved, not in shared folders.
- Use a password-protected device with full-disk encryption (BitLocker on Windows, FileVault on Mac).
- Don't send recordings through messaging apps or unencrypted email.
- Delete copies too, including those in your downloads folder and trash.
Checklist
- Consent form covers recording, transcription, storage, retention and withdrawal
- Recording made with cloud sync turned off
- Transcription done locally or with an approved service
- Transcript reviewed and de-identified
- Linking key stored separately
- Recording destroyed at the end of the retention period
Frequently asked questions
Can I use an online transcription service for research interviews?
Often yes, if your IRB has approved it, participants were informed in the consent form and any required agreements are in place. Many universities keep a list of approved vendors. Ask before you upload anything.
Do I have to delete recordings after my thesis is done?
Follow the retention period in your approved protocol and consent form. Some institutions require keeping data for a set number of years, others ask you to destroy recordings once the transcript is verified.
Is removing names enough to de-identify a transcript?
Usually not. A combination of job, location and age can still identify someone. Generalize those details too.
How long does it take to transcribe an interview?
By hand, plan for four to eight times the length of the recording. With speech recognition, the transcription itself takes minutes depending on your device, plus time for review.
This guide provides general information and is not legal advice. Your institution's IRB, data protection office and approved protocol take precedence. Last updated October 2026.