How much does it cost to have an interview transcribed?

Human transcription in the UK typically runs from about £0.95 to £1.45 per audio minute depending on turnaround, with academic intelligent verbatim commonly around £1.00 to £1.05. Internationally the usual band is roughly 1.00 to 3.00 US dollars per audio minute. An hour of interview therefore costs roughly 60 to 180 in the local currency.

What does automated transcription cost?

Considerably less. Automated services generally charge in the region of 0.19 to 0.49 per audio minute, or a flat monthly subscription. What you get is a raw transcript that still has to be checked against the audio and corrected.

Is a transcription agency worth it for a dissertation?

For eight 60-minute interviews an agency costs several hundred to well over a thousand, depending on the rate. For most dissertations that is the most expensive route available. It is worth it when a strict convention is required, the recording quality is poor, or there is simply no time left.

Is it allowed to outsource transcription of research interviews?

Yes. A transcription agency is a processor in data protection terms, so you need an agreement under Article 28 GDPR and a mention in the participants' consent form. Methodologically outsourcing is unproblematic as long as you do the correction pass yourself.

How much time do I save doing it myself with software?

Manual typing takes five to ten times the audio length. With automatic transcription the raw text appears in minutes, leaving roughly 30 to 60 minutes of correction and anonymisation per hour of audio. Forty to eighty hours of work becomes four to eight.

Cost comparison

Interview transcription cost

The question usually arrives late and under time pressure. The interviews are done, the deadline is closing in, and typing them out is no longer an option. This page sets the three routes against each other with real market rates and works them through on a typical thesis.

The three routes

Route	Price	Your time
Type it yourself	0	5 to 10 hours per hour of audio
Transcription agency, human	UK about £0.95 to £1.45 per audio minute; internationally about $1.00 to $3.00	Proofreading, roughly 20 minutes per hour of audio
Software with automatic recognition	About 0.19 to 0.49 per audio minute, or a flat monthly fee	Checking against the audio, 30 to 60 minutes per hour

Rates from published price lists of UK and European providers, 2026, usually excluding VAT. Surcharges are normal for poor audio quality, more than two speakers, strong accents and rush turnaround. Figures in different currencies are shown as published and are not converted.

Worked through: eight 60-minute interviews

A common scale for a master's thesis, so 480 audio minutes. The figures below use European rates in euro so the comparison stays in one currency.

Route	Cost	Time
Type it yourself	0	40 to 80 hours
Agency, standard rate €1.45/min	about €700 excl. VAT	about 3 hours proofreading
Agency, academic rate €2.50/min	about €1,200 excl. VAT	about 3 hours proofreading
Automated service, €0.30/min	about €145	4 to 8 hours
Nodl, two months on Starter	€58	4 to 8 hours

The same amount of audio, four routes. The top two bars are per-minute service prices; the bottom two additionally require four to eight hours of your own correction time.

The spread between the most and least expensive paid option is a factor of twenty. For most dissertations this is the single largest cost decision that gets made at all.

When each route is the right one

Typing it yourself

Sensible for one or two short interviews, or where the form of speech is itself under study and a detailed convention such as Jefferson notation is required. Then the typing is already part of the analysis rather than pure labour.

From about five hours of material it becomes uneconomic. Forty hours of typing is a full working week that is missing from your analysis.

Transcription agency

The right call with poor recording quality, strong accents, more than three speakers, or where a strict convention is demanded. Also when time has simply run out, because an agency delivers to a schedule.

Bear in mind: the agency is a processor. You need an agreement under Article 28 GDPR, and your participants' consent form has to state that an external service is involved. If you did not provide for that, you cannot simply outsource after the fact.

Software with automatic recognition

For the majority of qualitative projects this is the most economical route, because intelligent verbatim is sufficient there. The correction pass remains necessary and is not optional, because automatic recognition reliably misses technical terms, proper nouns and negations. One overlooked “not” inverts a finding.

The selection criteria are less about price than about three other questions: are speakers separated, are there timestamps for checking, and where is the data processed?

The cost that appears on no price list

Interview recordings contain personal data, and on sensitive topics also special categories under Article 9 GDPR. A cheap provider that processes outside the EU or uses content for model training can cost you your ethics approval or your data protection sign-off. That is not a price in pounds or euro, but a price in weeks.

The questions to settle before choosing are in Interview transcription and the GDPR.

What Nodl costs and what is included

Nodl does not bill per minute but as a monthly subscription. The Starter plan is €29 per month. For a dissertation whose transcription phase typically runs four to eight weeks, that is €29 to €58 in total.

Included for the interview use case:

Upload existing recordings, not only speak in live
Speaker separation with colour coding in the transcript
A timestamp per passage that jumps straight to that point in the audio
Processing and storage in Germany, language models inside the EU
Content is never used to train models

Two limitations worth knowing before you decide: a single recording may be up to one hour long, so longer interviews have to be split first. And the free tier allows one export, after which a plan is required. Recording, transcribing and reading are unlimited before that, so you can test recognition quality on a real interview before any money moves.

Common questions

Should I declare in my thesis that I transcribed automatically?

You should. The approach belongs in the methods section along with the convention you chose and a note that a correction pass against the audio took place. Many departments explicitly permit automatic transcription but require disclosure. Ask your supervisor if in doubt.

Why is academic transcription more expensive?

Because a defined convention is followed: consistent speaker labelling, marking of inaudible passages, pauses, timestamps at fixed intervals. That is slower than free note-taking and requires familiarity with the particular system.

Can I combine an agency with software?

Yes, and with difficult material it is often the cheapest solution. Pre-transcribe automatically and send only the problematic recordings, those with poor quality or strong accents, to the agency. Some providers price the correction of AI transcripts below transcription from scratch.

What about free tools?

For dissertations they are a serious option, particularly locally running models on your own machine, because no data goes to third parties and no processing agreement is needed. The price is setup effort, compute time and usually no speaker separation. With free online services, read the privacy policy on whether content is used for training.