What is transcription? Meaning, types and worked examples

Author: Mert S. Kaplan · Last updated: November 3, 2026

What is transcription? It is the writing down of speech from a recording or a live conversation: an interview, a meeting or a hearing, set out as text the way it was said. The word has other uses, in genetics and in music for example, so this guide sticks to speech. It separates transcription from what it is often confused with and shows the main types on the same example.

Transcription vs dictation, translation and captions

These get mixed up in searches and in tools:

Term	What it means	Example
Transcription	Writing down speech in the language it was spoken	An interview typed out from a recording
Dictation	Speaking so a machine or person types what you say	Voice typing a note on your phone
Translation	Moving meaning into another language	A Turkish interview rendered in English
Subtitles and captions	Timed text shown with a video	An SRT file for YouTube
Transcript	The finished text of a transcription	The document you quote from

Dictation serves the person speaking. Transcription usually has to cope with other people, several voices and a recording nobody made with a machine in mind.

Try it on your own recording

No card needed. 30 minutes of transcription on the free plan.

Transcribe a recordingGo to the app

Three types, one recording

Here is the same short conversation, an invented exchange at a town council, written in three ways.

Verbatim: Every sound, repeat and unfinished sentence is written.

Well, I mean, we, we um announced that park last month actually. But the budget, the budget never came [laughs], so what can you do?

Clean: Fillers and repeats go, while the meaning and choice of words stay.

We announced that park last month actually. But the budget never came, so what can you do?

Summary: The content is reported in the writer’s own words and can no longer be used as a quote.

The official said the park was announced last month but could not be built because the budget never came.

Which you choose depends on what the text is for:

Type	Who uses it	Watch out for
Verbatim	Qualitative research, legal work	Takes longest, and hesitations count as data
Clean	Journalism, meetings, content	Tidy without changing the speaker’s words
Summary	Meeting notes, briefings	Not really transcription, but an interpreted text

For a table of marks and an example see transcription conventions, and for how much detail research needs see qualitative interview transcription.

By hand or with AI?

By hand, you stop and start the recording as you type. A common rule of thumb is several hours of typing for each hour of audio, and longer if the recording is poor. In return, your ear often handles names and context better than a model does.

AI transcription does the same work in far less time, with mistakes. How accurate Whisper-style models are varies a great deal by model and recording, and we shared our own numbers in our Whisper test. The sound approach combines the two: the machine writes the first draft and you correct it while listening.

Which type LiveType produces

LiveType gives you text closest to the clean type. Fillers are left out (in Turkish and English). For a verbatim transcript, add hesitations and laughter yourself. It does not summarise or translate.

It also lets you edit while it transcribes, so you can fix a mistake the moment you hear it without new text disturbing your correction. If you are working through an interview, continue with our interview transcription guide. To find the route for your source, see the audio to text converter page. The free plan has 30 minutes to try, and the rest is on the pricing page.

Mert S. Kaplan

Works in news, software and design; founder of LiveType. Worked as a reporter for many years. Has done SEO consulting, content editing and reporting for several online news sites, and has developed many pieces of software. Currently works in the press office of the Turkish Medical Association and provides SEO support to Elips Haber.