Example 1: verbatim
Verbatim records everything: fillers, repetitions, broken-off sentences, pauses and non-verbal sounds. You use it when you also analyse how someone says something, for example in conversation or discourse analysis.
[FICTIONAL EXAMPLE]
Interview R3, 14-03-2026, length 42:10
I = interviewer, R3 = respondent 3 (district nurse)
[00:04:12] I: And how, uh, what does a normal morning look like for you?
[00:04:16] R3: Well, uh (.) normal, I usually start at, at half past seven
with the handover. And then, yeah, then you already see who (.)
who needs extra care, so to speak.
[00:04:31] I: Mhm.
[00:04:32] R3: And then, uh, then I drive (2.0) actually I drive more than
I care, haha. [laughs] No, that is an exaggeration, but it, it
does feel like that sometimes.
[00:04:45] I: How many clients do you see on a morning like that?
[00:04:48] R3: Eight, nine? On a bad day [unintelligible 00:04:50] ten.
Example 2: clean (intelligent verbatim)
Clean follows the content exactly, but drops "uh", repetitions and short acknowledgements like "mhm". This is the common form for thematic analysis and for most theses. Note: the respondent's words remain, nothing is rewritten or summarised.
[FICTIONAL EXAMPLE]
Interview R3, 14-03-2026, length 42:10
I = interviewer, R3 = respondent 3 (district nurse)
[00:04:12] I: What does a normal morning look like for you?
[00:04:16] R3: I usually start at half past seven with the handover. Then
you already see who needs extra care.
[00:04:32] R3: After that I actually drive more than I care. [laughs] No,
that is an exaggeration, but it does feel like that sometimes.
[00:04:45] I: How many clients do you see on a morning like that?
[00:04:48] R3: Eight, nine? On a bad day [unintelligible] ten.
Not sure which form your programme requires? The differences are explained in detail at what is verbatim transcription.
The notation used in the example
| Notation | Meaning | Tip |
|---|---|---|
| I: and R3: | Interviewer and respondent 3 | Use codes instead of names. Keep the key separate and secure. |
| [00:04:12] | Timecode in hours:minutes:seconds | Per turn is enough. That way you find a quote back in the recording. |
| [unintelligible] | Cannot be understood | In verbatim add the timecode, so a second listener can check it. |
| (.) and (2.0) | Short pause, and a 2-second pause | Only in verbatim, and only if you analyse pauses. |
| [laughs] | Non-verbal sound | Useful in clean too if it changes the meaning, as with irony. |
| [name] | Removed identifying information | For example [employer] or [town]. |
Transcripts in your thesis appendix
- Ask your supervisor first. Some programmes want all transcripts in the appendix, others only an example or a reference to a secure folder.
- One appendix per interview. For example Appendix C1: Interview R1, with date, length and a code instead of a name at the top.
- Line numbers or timecodes. Then you can refer to them in the text: (R3, 04:32). In Word you turn on line numbers via Layout > Line Numbers.
- State how you transcribed. For example: "The interviews were transcribed automatically with Murmel and then fully replayed and corrected by the researcher." That belongs in your methods chapter.
Anonymising and consent
Before the interview, ask for written consent to record and to have the recording transcribed. Record what you use it for, who may see it and when the recording will be deleted. Your university often has its own form and rules for storing research data; follow them.
Anonymise in the transcript, not just in the thesis. Replace names with codes and remove details that still make someone recognisable: employer, role in a small team, home town, a striking event. Combinations make someone identifiable, even without a name.
With Murmel, audio and transcript are processed and stored in the Netherlands. Recordings are not used to train the model unless you turn that on, and you can have them deleted automatically after 1, 7, 30 or 90 days or a year. More for researchers at transcription for research.
How to make this with Murmel
- Upload the recordingDrag the file into the Studio. A one-hour interview is usually ready in under a minute, with speakers separated.
- Rename the speakers to codesTurn "Speaker 1" and "Speaker 2" into I and R3 straight away. Then no name appears anywhere in your export.
- Replay and correctMurmel delivers a clean transcript. Click a word to hear the audio at that spot. For verbatim, add fillers and pauses back where your analysis needs them.
- Export with timecodesWhen exporting, choose Blok per spreker met tijdcodes (block per speaker with timecodes) as the text format. Download as Word or TXT and paste it into your appendix.
For a thesis with ten one-hour interviews you pay €14.50 incl. VAT with prepaid credits. Credits are valid for six months. See pricing and the workflow at transcribe interviews.
Frequently asked questions
Do I have to transcribe my interviews verbatim?
Usually not. Clean is common for thematic analysis. You need verbatim if you analyse how people speak. Your method and your supervisor decide.
Can I use an automatic transcript for my thesis?
At most programmes yes, as long as you check it and state in your methods chapter how it was made. Check your programme's rules and whether the storage location fits your data management plan.
How long does it take to type up an interview yourself?
Manually, an hour of audio often takes four to six hours of typing. With an automatic transcript mostly reviewing and correcting remains.
Can Murmel handle English interviews?
Yes, Dutch and English, also mixed in one conversation. Not other languages.