How to Transcribe Interviews: Methods, Options and Costs
What a good interview transcript needs, how to transcribe interviews by hand, with a service, a cloud tool or self-hosted software, what to watch for in a dissertation or academic research, and what transcription costs.
An interview is quick to record. Turning it into text you can quote, code and argue from takes considerably longer than the conversation itself. Whether it is for a master's thesis, a dissertation, a research project, a piece of journalism or an internal survey, anyone who needs to transcribe interviews runs into the same questions. What does the transcript have to do, who writes it, what does it cost, and where do the recordings end up along the way?
This guide works through the answers in order. It sets out what makes an interview transcript usable, compares the ways one gets made with their pros and cons, and covers what to check when you transcribe interviews for academic research. It closes with a short walkthrough of transcribing interviews with Scriber.
What a Good Interview Transcript Needs
A transcript is good when you can work from it without listening to the whole recording again. Three properties decide that.
Speaker separation. Every turn has to be attributed to the person who said it, whether interviewer or interviewee, and in a focus group to every individual participant. Without that attribution a transcript is one continuous block of text, and you reconstruct from memory who said what. For a quotation in a paper, attribution is the foundation.
Timestamps. Time references at each change of speaker, or at regular intervals, make a passage findable in the recording. That matters when you want to listen again because tone, pauses or an unclear phrase affect the interpretation, and when a supervisor or reviewer wants to check the passage.
Verbatim or clean. A verbatim transcript records what was said, including filler words, false starts, repetitions and dialect. A clean (or intelligent verbatim) transcript turns speech into readable written language and drops fillers and slips. Which one is right depends on your method. Conversation or discourse analysis usually needs the verbatim version, while a content-focused qualitative analysis is often fine with a clean one. Many disciplines work with fixed transcription conventions. Decide on yours before the first interview and apply them consistently across every transcript. The common systems and how to choose between them are covered in Transcription Conventions: Choosing a System for Your Interviews.
Then there is the format. A transcript you have to retype or reformat before it fits your analysis software or word processor costs you the time you were trying to save. Some conversations do not need a full transcript at all. When minutes or notes will do is explained in Transcript vs. Minutes: What's the Difference?.
Four Ways to Transcribe an Interview
An interview transcript is made in one of four ways. Which one fits depends on volume, budget, time, and how confidential the material is.
Transcribing by hand
You play the recording, stop, type and rewind until every sentence is right.
- Pros: no cost beyond your time, full control over the transcription conventions, and the recording never leaves your computer. Transcribing your own interviews also means you get to know the material very well.
- Cons: it takes many times the length of the interview, and the work grows with every additional conversation. With many interviews, or focus groups with several voices, it quickly becomes the bottleneck of an entire project.
Using a transcription service
You send the recordings to a provider who has them transcribed by people, by software, or by a combination of both.
- Pros: the work is no longer yours, and many services offer a choice of transcription conventions, speaker separation and timestamps.
- Cons: billing is usually per minute of audio, so the cost rises with every interview. The price typically depends on audio quality, the number of speakers, the accuracy you need and the turnaround time. Your recordings go to a third party. With personal data in them, that usually requires a data processing agreement and a basis in your participants' consent forms.
Using a cloud transcription tool
You upload the recording to an online service that transcribes it automatically, often within minutes.
- Pros: fast, needs no infrastructure of your own, and usually comes with automatic speaker separation and export to common formats.
- Cons: billing follows usage, per minute or as a subscription with a quota. The audio is processed on the provider's servers. Where exactly, which sub-processors are involved and how long data is kept are things you need to check before the first upload. For confidential research interviews, that is often where this route stops.
Using self-hosted software
The transcription software runs on a server belonging to your institution or organisation. Recordings are processed there and never leave your own network.
- Pros: automatic transcription with speaker separation, without sending audio to a third party. The cost does not depend on how many hours you transcribe, and data protection questions can be answered on your own systems.
- Cons: it needs a suitable server and someone to look after the installation. For one person with three interviews it is rarely worth it. For a research group, an institute or a project with a steady flow of interviews it is much more likely to be. More on the page about self-hosted transcription on your own servers.
What Interview Transcription Costs
There is no honest single figure. Transcription prices vary considerably by provider, pricing model and volume. The cost comes about in different ways depending on the route:
- By hand, you pay in working time, and that grows with every interview.
- With a service or a cloud tool, you usually pay per minute of audio or through a usage-based subscription. The bill rises with volume.
- With self-hosted software, you pay for a licence and for the server it runs on. More interviews change little about either.
How to estimate your monthly audio minutes, and which costs never appear in a per-minute rate, is covered in Transcription Cost per Minute: What Actually Drives the Price. How Scriber is licensed is on the Scriber pricing and licensing page.
Interview Transcription for a Dissertation or Academic Research
For a dissertation or any other academic work, a third question joins time and money, which is where the recordings are allowed to go. Interviews almost always contain personal data, often of a sensitive kind, such as health, ethnic origin, political opinions or employment. Academic transcription therefore comes with a few points to settle before the first interview. The whole process, from planning your time to what supervisors expect, is covered in Transcribing Interviews for a Dissertation or Thesis.
- Consent: the consent form should describe who processes the recordings, where they are stored and when they are deleted. If you plan to use a transcription service, that has to match what your participants signed. What the form should cover is set out in Interview Recording Consent and GDPR: What to Cover.
- Data processing: if you hand recordings to a service or a cloud tool, it is usually acting as a data processor. Check the data processing agreement, the processing location and the list of sub-processors before you upload the first file.
- Keeping audio in-house: many universities and ethics committees prefer raw data not to leave their own infrastructure. Transcription on a university or institute server meets that without giving up automatic transcription.
- Pseudonymisation: replace names, places and other identifying details in the transcript with codes or pseudonyms before you share or quote passages, and keep the key separately.
- Who to ask: settle the exact wording for your study with your university's data protection officer and, where required, your ethics committee.
What this looks like for a whole research group or institute is described on the page about interview transcription for research.
How to Transcribe Interviews with Scriber, Step by Step
Scriber is self-hosted transcription software. It runs on a server belonging to your institution, and recordings are processed only there. From recording to transcript, the process looks like this:
- Bring in the recording. Upload the interview file to Scriber, or record directly in the browser. With many interviews, put the files in a watched folder on the server, an S3-compatible bucket or a network share instead. Scriber then transcribes whatever arrives there without anyone opening the application.
- Get a speaker-separated transcript. Scriber transcribes the recording and attributes every turn to the person speaking, including in focus groups with several voices.
- Review and pseudonymise. Read the transcript against the recording, check the speaker attribution against your notes, and replace identifying details according to your transcription conventions.
- Export. Export the transcript as text, Markdown, JSON, Word or PDF and bring it into your analysis software or your thesis.
- Find it again. Every transcript sits in a searchable library, so you can search all your interviews at once for a passage.
If you want to see what a speaker-separated transcript from Scriber looks like first, try the free hosted demo at scriber.determin.de with a recording that is not confidential. Real interviews belong in your own installation. For a quote for your research group, institute or organisation, write to [email protected].
Frequently Asked Questions
How do I transcribe an interview?
You can type it up by hand, send it to a transcription service, use a cloud tool for automatic transcription, or run self-hosted software on a server of your own. Which route fits depends on how many interviews you have, your budget, your time and how confidential the recordings are. Whichever you choose, decide up front whether you need a verbatim or a clean transcript, and look for speaker separation and timestamps.
How much does interview transcription cost?
It depends on the method and the provider, so there is no universal figure. Transcription services and cloud tools usually bill per minute of audio or through a usage-based subscription, so the cost rises with every interview. By hand, you pay in working time. Self-hosted software such as Scriber is licensed per user, with no per-minute billing.
Can I use a transcription service for my dissertation interviews?
Usually yes, as long as your participants' consent forms cover it and you have a data processing agreement with the service. Also check where the provider processes the recordings and how long it keeps them. Many universities prefer raw data not to leave their own infrastructure. Settle the exact arrangement with your university's data protection officer.
What should an interview transcript look like?
Every turn is attributed to the person speaking, timestamps make passages findable in the recording, and the whole transcript follows the same transcription conventions throughout, verbatim or clean. To be useful, it should go into your analysis software or word processor without retyping.
Ready to See Scriber on Your Own Servers?
Tell us about your infrastructure and workload, and we will answer with a concrete recommendation.