Interview Transcription with Speaker Labels — for Journalists, Researchers, and HR

Transcribe an Interview in 4 Steps

Same workflow whether you recorded on a phone, headset, Zoom, or DSLR camera.

Cost reference: a 1-hour interview costs about $0.30 on the Starter plan ($2/mo, 200 minutes) — full comparison against pay-per-file and human interview transcription services on the cost calculator.

Who Uses Interview Transcription

Same tool, different formatting expectations. Match your workflow:

How to Format an Interview Transcript

The formatting decisions that matter for interview transcripts — and what verbatim vs clean-copy conventions actually change:

Rule of thumb: use verbatim for research, legal, and testimony; clean copy for journalism narrative, HR summaries, and internal knowledge base entries. The transcript is editable in-browser — you can toggle between conventions post-hoc without re-transcribing.

The Three Transcription Levels — True Verbatim, Intelligent Verbatim, Clean

Same recording, three different outputs. Choose the level that matches your job-to-be-done:

Which one? Qualitative researchers → intelligent verbatim (most common in published qualitative research per Silverman's Interpreting Qualitative Data and Bryman's Social Research Methods). Journalists and podcasters → clean copy. Legal, court, or discourse analysis → true verbatim. VexaScribe produces intelligent verbatim by default; true verbatim mode retains filler words and non-verbal markers; clean copy is generated on the second pass.

How to Do Verbatim Transcription — the Rules

True verbatim transcription follows conventions from qualitative research methodology and legal transcription. The standard notation:

• Filler words preserved: um, uh, er, like, you know, I mean — all kept
• False starts / self-repairs preserved: “I was going—I wanted to say”
• Non-verbal vocalizations bracketed: [laughter], [sigh], [cough], [throat clear]
• Pauses marked with duration: [pause 2s], [long pause], [silence]
• Overlapping speech: [overlapping speech] or Jefferson-style brackets [[ ]]
• Emphasis marked: WORD in caps for stressed emphasis; underline in DOCX export
• Inaudible / unclear: [inaudible], [unclear at 12:34], [crosstalk]
• Non-lexical / paralinguistic: retain hedge markers (mm-hmm, uh-huh) as they carry meaning in interviews
• Speaker labels consistent: INTERVIEWER, PARTICIPANT (or P1, P2, P3 across a corpus for anonymization)

These conventions are documented in qualitative research methods textbooks (Silverman 2020, Bryman 2016) and Jefferson's conversation-analysis notation system (widely used in discourse analysis). For legal transcription, add certification headers per jurisdiction — see deposition transcription for that workflow.

Export Workflows for NVivo, MAXQDA, Atlas.ti, and Dedoose

The 4 major qualitative analysis tools each have their own import requirements. Here's the honest export recipe for each:

All four tools accept the standard intelligent-verbatim DOCX output from VexaScribe. The differences are speaker-prefix format and timestamp syntax. Consistency across your interview corpus matters more than which tool you use.

What the Interview Transcript Looks Like

Clean-copy example from a qualitative interview. Note consistent speaker labels, second-precision timestamps, and paragraph breaks per turn:

[00:00:00] INTERVIEWER: To start, could you tell me a little about your role and how long you've been in it?

[00:00:07] PARTICIPANT: Sure. I've been in product design for about eight years — the last four at my current company. I lead a team of six.

[00:00:16] INTERVIEWER: And when I asked earlier about your day-to-day, you mentioned feeling stretched. Can you walk me through what a typical week looks like?

[00:00:23] PARTICIPANT: Yeah, so — [pause 2s] — I think the hardest part is context-switching. Monday is usually planning, Tuesdays and Wednesdays are heads-down design work, and then Thursdays and Fridays I'm mostly in meetings with engineering and stakeholders.

[00:00:41] INTERVIEWER: When you say "context-switching is the hardest part," what specifically makes it hard?

Verbatim mode retains "yeah, so — [pause 2s] —" and the filler-word pauses shown here. The clean-copy mode above is the default.

Qualitative Interview Transcription for Research

Qualitative research has specific requirements that generic transcription doesn't always meet. What's different:

• Coding-friendly export: NVivo, ATLAS.ti, and MAXQDA all accept plain-text or DOCX with line numbers. Export line-numbered DOCX for direct import.
• Consistent speaker attribution: a research corpus of 20 interviews needs the same speaker labeling scheme across all files. SPEAKER 1 / SPEAKER 2 stays consistent; rename to role-based tags (P1, P2, INT) if your codebook expects it.
• Verbatim requirement for discourse analysis: if you're coding conversation dynamics (turn-taking, overlap, hedging), verbatim mode is not optional. Filler words and pauses are the data.
• Reflexivity notes stay separate: researcher memos and reflexivity notes shouldn't live in the transcript — keep them in your qualitative software's memo function or a separate log.
• Anonymization pass: for IRB-sensitive material, run the exported transcript through a redaction step (or use verbatim search-and-replace) before importing to coding software. AI transcription does not automatically redact names.

AI Interview Transcription vs Human Transcription

The honest cost/accuracy comparison, without marketing polish:

The middle path most researchers use in 2026: AI transcription (~$0.30/hr) plus a 20–30 minute manual editing pass gets you to ~99% accuracy at 1/100th the wall-clock cost of full human transcription. For step-by-step process guidance, see how to transcribe an interview.

Free Interview Transcription — What 30 Minutes Gets You

Honest version: 30 minutes free covers one standard interview session. It doesn't reset monthly — it's a trial. For ongoing free interview transcription, the actually-free options are:

• OpenAI Whisper installed locally — free forever, offline, unlimited. Requires Python + a decent GPU. Best for confidential interviews where cloud upload creates compliance risk.
• Otter.ai free tier — 300 min/month but capped at 30 min per conversation. Fine for short structured interviews, breaks on 60-min narrative ones.
• TurboScribe free tier — 3 files/day, 30 min cap each.

Not an Interview? Start on the Right Page

Related Guides