What an interview transcript looks like, and how to write one

An interview transcript is a written record of what was said in an interview, laid out speaker by speaker. Most use intelligent verbatim: every word, minus the fillers. A usable transcript carries a short header, a label for each speaker, timestamps on a fixed rule, and tags for anything unclear.

What is an interview transcript?

interview transcript A written, speaker-by-speaker record of an interview, produced so the conversation can be read, quoted and analysed instead of replayed. Scribbr defines transcribing as converting speech to text word for word.

People also call it a transcribed interview, a transcript from an interview, a transcript of an interview or an interview transcription. Same document, same job. It sits next to the recording rather than replacing it, because the audio settles any dispute about wording.

Verbatim, intelligent verbatim, edited

StyleWhat it keepsWhat it dropsWhen to use it
VerbatimEvery word, plus pauses, laughter, stuttering and hesitations such as "uh".Nothing. Readability is the price.When how something was said matters. Scribbr names legal work and some research.
Intelligent verbatimEvery word that carries meaning. Grammar and broken sentences may be tidied.Fillers like "uhm", "yeah" and "you know". With them go emotion, pauses and hesitation.The default for most interviews. The most readable option that is still word for word.
EditedThe substance, summarised and edited down from an intelligent verbatim draft.Irrelevant sentences, where cutting them does not change the meaning.Publication and internal write-ups, where the reader wants the point, not the record.

An interview transcript example you can read

Study ID: GR-2026-04
Participant ID: P07 (pseudonym: "Dana")
Date: 2026-04-18
Interviewer: S. Okafor
Session type: Remote, 1 to 1, 32 minutes
Recording file: GR2026-04_P07.m4a
Transcription style: Intelligent verbatim
Timestamps: Every speaker change
Consent: Verbal consent recorded at 00:00, written form on file
Speaker key: I = interviewer, P = participant

[Begin Transcript]

[00:00] I: Before we start, can you confirm you are happy for this to be recorded?

[00:06] P: Yes, that is fine.

[00:09] I: Thank you. Tell me how you plan a week of deliveries now.

[00:14] P: It starts on Sunday night. I print the route list, then I go through it with a pen, because the system orders them by postcode and that is not how the roads work. [laughs] It takes me about an hour.

[00:38] I: What happens when something changes on the day?

[00:42] P: Then it is phone calls. The depot rings me, I ring the driver, and [inaudible 00:04:12] nobody updates the sheet until the evening.

[01:05] I: So the sheet is behind for most of the day.

[01:09] P: Most of the day, yes.

[End Transcript]

The same exchange, verbatim and intelligent verbatim

VerbatimIntelligent verbatim
P: So, um, yeah, I mean, it starts on Sunday night, [pause 3s] Sunday night, yeah.P: It starts on Sunday night.
P: I print it out and, and I go through it with a pen, you know, because, uh, the system orders by postcode.P: I print it out and go through it with a pen, because the system orders by postcode.
I: [both speaking, unclear] sorry, go on. P: No, no, you go.I: Sorry, go on. P: No, you go.
P: It takes, I would say, about an hour? ((laughs)) Maybe more.P: It takes about an hour. [laughs] Maybe more.

What every interview transcript header should carry

How to properly transcribe an interview in 6 steps

  1. Ask before you record, and capture the yes Get permission, then ask again once recording has started so the consent is on the recording itself. Otter's published workflow also opens with permission, and its template carries a consent field. "Before we start, can you confirm you are happy for this to be recorded?"
  2. Choose the style and the timestamp rule Decide now whether this is verbatim, intelligent verbatim or edited, and how often timestamps appear. Write both into the header. GoTranscript's template treats this as a field you fill before transcribing. Common rules: every speaker change, every 2 to 3 minutes, or every 5 to 10 minutes.
  3. Get a first draft Either run the audio through software or type it out yourself. Software needs clean input to be worth it. If the audio is poor, dictating it or typing it by hand may be the only route.
  4. Add speaker labels Apply one convention throughout. Initials, role tags, full words or participant codes all work. Rev's rule is a full name on first mention, then first name, initials or title. I: and P:, or INTERVIEWER: and PARTICIPANT:, or MOD, P1, P2 for a group.
  5. Fix the draft against the audio and tag the gaps Play it back and correct as you read. Where you cannot hear a word, tag it rather than guess. Where you added a word so the sentence reads, mark the addition in brackets. "I showed him that this option [raising prices] would be beneficial."
  6. Store the transcript with the recording Keep them together under the same file reference, then move on to analysis. Coding is one common next step: link keywords to answers so you can compare what different people said.

The time budget

Rev states that the average person can transcribe 1 hour of audio in about 4 hours. Software collapses that to minutes, but only on clean audio. Paid interview transcription services exist; turnaround and accuracy are claimed per vendor.

Tags for everything you cannot transcribe cleanly

SituationCommon tagHow it looks in a line
A word or phrase you cannot hear[inaudible] or [inaudible 00:12:43]P: We moved the whole [inaudible 00:12:43] over to the new depot.
Two people talking at once[overlap] or [both speaking, unclear]I: [both speaking, unclear] sorry, go on.
A noticeable silence[pause] or [pause 3s]P: It was, [pause 3s] honestly, it was a mess.
Laughter or another non-verbal sound[laughs], ((laughs)), [cough]P: It takes about an hour. [laughs]
A word you added so the sentence readsSquare brackets around your additionP: I told them [the depot team] on the Friday.
A word said with emphasisItalicsScribbr's example: Increasing prices is needed for profitability.

How to format an interview transcript, and what to avoid

Do

Do not

Where interview transcription software slips

Software for transcribing interviews is only as good as its input. Scribbr's conditions are a noise-free background, no over-talk, clear accents and good microphones. Miss those and you are editing more than you would have typed. Vendors publish their own accuracy figures: Rev states 99 percent or better for its human service, Otter states 95 percent or better in clear audio. Both are claims about their own product.

So re-listen selectively. Names, numbers and technical terms are where any interview transcriber fails first, machine or human, followed by any stretch where two people talked over each other. Zoom labels a speaker it cannot identify as "Unknown Speaker", a useful reminder that speaker attribution is a guess too. For choosing between tools, see our note taker comparison rather than a list here.

A transcript you cannot trust is worse than no transcript. Tag the gap instead of guessing at it.

When the interview was a hiring interview

Then the transcript is not only a research artefact. Under 29 CFR 1602.14, any personnel or employment record an employer makes or keeps, including records having to do with hiring, must be preserved for 1 year from the date the record was made or from the personnel action, whichever is later. The EEOC's own employer guidance states the 1 year rule plainly.

Once a discrimination charge is filed, the same regulation requires all personnel records relevant to the charge to be kept until final disposition, and "relevant" expressly reaches the application forms and test papers of other candidates for the same position. These are US federal minimums for covered employers. Set a retention period deliberately, and take advice for your jurisdiction.

AI interview transcription, and what it produces

The transcript card after an AI interview: speaker labels down one side, MM:SS timestamps, the recording alongside.

On The Cognitive the interview itself is the AI interview, so nobody has to make the transcript. It is generated once the interview ends, labelled by speaker, and stored with the recording and the scorecard against the role. It does not join or transcribe your own Zoom, Teams or Meet calls.

What to remember

Where the transcript turns into a decision

A scorecard built from the same interview: each criterion out of 5, an overall score out of 100, and a recommendation.

A transcript is a record, not a judgement. The scorecard sits beside it: every criterion scored out of 5, an overall out of 100 weighted from the role's evaluation criteria, and a "Verified in this interview" section marking resume claims verified, refuted or unclear, with the evidence behind each one.

Related reading and free tools

Best AI note taker for recruiters · Interview feedback examples · Video interview software · AI interviewer guide · Interview kit

Frequently asked questions

What is an interview transcript?

It is a written record of an interview, set out speaker by speaker, so the conversation can be read, quoted and analysed rather than replayed. Transcribing means converting speech to text word for word. Asking what is a transcription of an interview gets the same answer: transcript, transcription and transcribed interview all name one artefact, which normally lives alongside the recording rather than replacing it.

How do you transcribe an interview?

Ask permission and capture the yes on the recording. Choose your style and timestamp rule and write them into the header. Produce a first draft by software or by hand, add speaker labels, then play the audio back and correct the draft, tagging anything you cannot hear. Store the finished transcript with the recording under the same file reference.

How do you write a transcript of an interview?

Work top down. The format for an interview transcript starts with the header: interviewer, interviewee or pseudonym, date, location or platform, recording file, style, timestamp rule and consent note. Then label every turn with the same speaker convention, place timestamps on the rule you chose, and use one tag set for unclear audio, overlaps, pauses and non-verbal sounds. That is how to write an interview transcript someone else can rely on.

What do interview transcripts look like?

The standard interview transcript layout is a short metadata header, then alternating labelled turns with a timestamp on each and a blank line between speakers. Labels vary: initials with a colon, role tags such as I and P, full words like INTERVIEWER and PARTICIPANT, or codes like MOD, P1 and P2 for groups. Some use the simplest interview transcript format of all, a Q: A: example where Q: marks the question and A: the answer.

How long does it take to transcribe a 15 minute interview?

By hand, budget about an hour to produce 15 minutes of a transcribed interview. Rev states the average person transcribes 1 hour of audio in roughly 4 hours, which scales to about 60 minutes of work for a 15 minute recording. Software turns the first draft into minutes, but you still need to check it, and Scribbr notes the audio has to be clean for that draft to be worth having.

How does Zoom transcription work?

Zoom can transcribe the audio of a cloud recording automatically. Once processed, the transcript appears as a separate VTT file beside the recording, can be shown as closed captions and can be edited. It includes timestamps per section and speaker identification, labelling anyone it cannot identify as "Unknown Speaker". It requires a Pro, Business, Education or Enterprise account with cloud recording and audio transcription enabled, and Zoom documents 19 languages.

Can an AI interview assistant transcribe both interviewer and candidate?

On The Cognitive, yes, because the AI runs the interview rather than listening in on one. Both sides are captured, each line carries a speaker label and an MM:SS timestamp, and the transcript is generated after the interview ends, next to the recording and the scorecard. It does not join or transcribe your own Zoom, Teams or Meet calls.

How long should we keep an interview transcript?

For US hiring, 29 CFR 1602.14 sets a floor of 1 year from the date the record was made or from the personnel action, whichever is later. If a discrimination charge is filed, relevant personnel records must be kept until final disposition of the charge or action. Those are minimums for covered employers. Set your own period deliberately and take advice for your jurisdiction.

What is a transcript for a job?

Usually something else entirely. When an employer asks for "your transcript", they normally mean an academic transcript from a school, college or university showing the courses you took and the grades you received. An interview transcript is the written record of an interview conversation. The two documents share a word and nothing else.

Explore: Best AI note taker · Interview feedback examples · Video interview software · AI interviewer guide