Interview transcription Hours of interviews, transcribed while you analyse.
For a master’s thesis, a PhD, a qualitative study or a news story: Butler transcribes your interviews with separated speakers and time stamps, you correct in place and export to Word — without the recordings going to a third-party transcription service.
- Interviewer and respondents separated
- Time stamps on every passage
- Word export for MAXQDA, NVivo, ATLAS.ti
Workflow
From the recorder to a transcript you can code.
- 1
Upload the recording
MP3, WAV, M4A or MP4 from a recorder, a phone or a video call — or record straight into Butler.
- 2
Set the language and speakers
That gives the best first transcript, especially with two voices that sound alike.
- 3
Pseudonymise
Rename a speaker once — “I” and “R1” instead of real names — and Butler changes it throughout.
- 4
Correct
Click a sentence, hear the recording at that point, fix it. Names and technical terms are cleaned up fast.
- 5
Export
As Word or PDF with speakers and time stamps — then import into your analysis software like any Word document.
- 6
Analyse
Optionally put every interview in a knowledge base: ask across the whole corpus, every answer with its location.
Transcription conventions
Verbatim or clean — your call.
Qualitative research works to set transcription conventions. Butler gives you both basic forms from the same recording; you apply your department’s rules while correcting.
Verbatim
The raw transcript
As close to the speech as recognition hears it — the basis for a verbatim transcript.
Clean
The AI version
Lightly smoothed wording, a speaker’s fragments merged — readable for thematic analysis.
Summarised
The report
With your own “research interview” template: themes, key statements, quotes with time stamps.
Check every transcript against the recording before you quote from it — true of any automatic transcription.
Research data
Interview data does not belong in just any online service.
Interviews are personal data, often sensitive: health, work, political views. Consent forms, data management plans and ethics boards usually require you to know where the recordings are processed.
Butler transcribes on its own GPUs — no third-party transcription API, no training on your recordings. With the complete version everything stays in your building; which version computes where is spelled out on the pricing page.
Questions about interview transcription
How do I transcribe an interview with AI?
Upload the recording, set the language and the number of speakers, and let Butler transcribe it. Then name or pseudonymise the speakers, correct the text against the audio and export to Word.
How accurate is it?
Very accurate on a good recording; names, technical terms and strong accents need corrections. That is why the audio sits beside the text and every passage has its time stamp.
Can I use the transcripts in MAXQDA, NVivo or ATLAS.ti?
Yes: export to Word and import the document into your analysis software like any other text document.
May I use AI to transcribe interviews for my thesis?
That depends on your consent forms and your university’s rules. What usually matters is where the data is processed — with Butler on its own servers, never at a third-party transcription service.
Which languages?
Every language Whisper large-v3 knows — including interviews that switch between two languages, once you set the main one.
Try it on one interview.
Send us a recording you are allowed to share — we show you the transcript Butler makes of it.