How to write a transcript: format, examples and a template

October 30, 2026 · 5 min read

A good transcript is easy to read, shows who said what, and lets anyone find a quote in the original recording. Getting there takes three decisions: how literal the text should be, how speakers are marked, and where timestamps go. This guide covers each one, with examples and a template you can copy.

Step 1: Choose the transcript style

There are three common styles. Pick one before you start and stick to it.

Style	What’s kept	Use it for
Verbatim	Every word, including “um”, “uh”, repetitions, false starts, and often pauses and laughter	Legal records, some research methods (e.g. conversation analysis), anything where how something was said matters
Clean verbatim	Everything that was said, minus filler words, stutters and false starts	Interviews, articles, research quotes, meeting records. The most common choice
Edited	The content, rewritten into correct sentences	Published Q&As, blog posts, reports

Here is the same answer in each style:

Verbatim:
Um, so I- I never, uh, planned to open a shop. The stall was, like, supposed to pay for my car.

Clean verbatim:
I never planned to open a shop. The stall was supposed to pay for my car.

Edited:
She never planned to open a shop; the stall was only meant to pay for her car.

If you’re not sure, use clean verbatim. It’s faithful to what was said and still pleasant to read.

Step 2: Add a header

Start the document with the basic facts, so the transcript still makes sense months later:

Interview with Maya Torres, founder of Torres Bakery
Interviewer: Sam Lee
Date: 21 October 2026 · Recorded on Zoom · Length: 42:16
Transcript style: clean verbatim

Step 3: Mark who is speaking

Start a new paragraph every time the speaker changes, and begin it with the speaker’s name or label in bold or followed by a colon.

Write the full name the first time, then a short label: “Maya Torres:” first, “Maya:” or “MT:” after that.
For interviews, “Interviewer:” and the interviewee’s name is enough.
For research, you may need anonymised labels such as “P1:”, “P2:” (participants) and “I:” (interviewer).
If you can’t tell who is speaking, write “Speaker 1:”, “Speaker 2:” and keep it consistent.

Step 4: Decide where timestamps go

Timestamps let a reader jump to the exact moment in the recording. Common choices:

At every speaker change: best for interviews and quotes.
At the start of each paragraph: good for lectures and monologues.
Every 30 seconds to 2 minutes: enough for meetings and long recordings.

Use one format throughout, usually [mm:ss] for recordings under an hour and [hh:mm:ss] for longer ones.

Step 5: Mark what you can’t hear

Never guess silently. Mark problems so the reader knows where the text is uncertain:

Notation	Meaning
`[inaudible 12:04]`	A word or phrase you can’t make out, with the time so someone can check it
`[unclear: Brennan?]`	Your best guess at a word, marked as a guess
`[crosstalk]`	Several people talking at once
`[laughs]`, `[pause]`, `[phone rings]`	Non-speech sounds, if they matter for the meaning

In clean verbatim, include only the sounds that change the meaning (a laugh that shows a remark was a joke, for example).

Transcript template

Copy this into Word or Google Docs and fill it in:

[Title: who, what, why]
Interviewer / Participants: [names or labels]
Date: [date] · Place or platform: [Zoom, phone, in person] · Length: [hh:mm:ss]
Transcript style: [verbatim / clean verbatim]

[00:00] Interviewer: Thanks for joining. Can you start by telling us how the business began?

[00:08] Maya Torres: Sure. It started as a weekend stall at the farmers' market...

[01:42] Interviewer: What made you decide to open a shop?

[01:47] Maya: Honestly, the customers did. [laughs] They kept asking where they could find us on weekdays.

How to type up a transcript faster

Typing a transcript from scratch commonly takes 3–4 hours per hour of clear audio. Two things cut that dramatically:

Start from an AI draft. Upload the recording to a transcription tool, then correct it instead of typing it. With TranscribeNest you can upload a file, get a transcript in minutes with a timestamp at the start of each paragraph, and download it as a Word document. It doesn’t label speakers yet, so you add the names in step 3 above.
Check while listening. On the transcript page, clicking a line plays that moment of the recording, and you can search for names and terms to check them all in one go. Names, numbers and technical words are where speech recognition is most often wrong.

If you’re transcribing by hand, use a player with keyboard shortcuts for pause and rewind, and slow the audio to 0.7–0.8× so you can type along.

One style (verbatim or clean) used throughout
Header with names, date and length
Every speaker change starts a new paragraph with a label
Timestamps in one consistent format
Names and terms spelled the same way everywhere
Unclear passages marked, not guessed
One last read-through while listening to anything that looks odd

FAQ

What’s the difference between a transcript and meeting notes? A transcript records what was said, in order. Notes or minutes summarise decisions and actions. Many people make the transcript first and write the summary from it.

Should I correct grammar in a transcript? In verbatim and clean verbatim, no: keep the speaker’s words, only removing filler in clean verbatim. Correcting grammar is what turns it into an edited transcript.

How do I quote from a transcript? Quote the words exactly as transcribed and keep the timestamp in your notes, so you or an editor can check the quote against the recording. See also how to transcribe an interview.