Overview
Different analytic approaches need different transcripts. Conversation analysis depends on pauses, overlaps and intonation; discourse analysis often needs hesitations and false starts; thematic analysis of experiences can usually work from a clean verbatim transcript that removes filler words without changing meaning. Choosing the style before transcription starts avoids re-doing work later.
We agree a transcription specification with you: verbatim style, speaker labeling, timestamp frequency, notation for inaudible passages and non-verbal events, and how names and other identifiers should be treated. Where automated speech recognition is used to produce a first draft, a person reviews the full recording against it; you can also ask for a fully human process if your ethics approval requires it.
Confidentiality is handled as part of the method, not an afterthought. Recordings are transferred and stored securely, access is limited to the people working on your files, and files are deleted after delivery on the schedule you specify. We can sign a confidentiality agreement or work under your institution's data processing terms.
Who it’s for and the problems it solves
This is for you if you are…
- Students and researchers with interview, focus group or oral history recordings awaiting analysis
- Researchers whose ethics approval requires specific confidentiality or de-identification steps
- Teams running multilingual fieldwork who need transcripts in the original language, translations or both
- Evaluation and market research teams needing timestamped transcripts for coding and quotation
- Researchers who tried automatic transcription and found it unreliable for accents, jargon or group discussions
Common problems we help with
- Automated transcripts mislabel speakers, mangle technical terms or miss crosstalk in focus groups
- You are unsure whether your analysis requires full verbatim or clean verbatim
- Your ethics approval says transcripts must be de-identified but you have no consistent procedure
- You have recordings in more than one language and need both transcription and translation
- Hours of recordings are delaying analysis and your deadline is fixed
What’s included — and what isn’t
Included
- Transcription in full (true) verbatim, clean verbatim or an agreed intermediate style
- Consistent speaker labels (for example I for interviewer, P01 for participant) and focus group speaker identification where audible
- Timestamps at set intervals, at each speaker change, or at inaudible passages
- Standard notation for inaudible or uncertain words, overlaps, pauses and non-verbal events
- Glossary of names, technical terms and acronyms supplied by you to improve accuracy
- De-identification or pseudonymization: replacing names, places and organizations with codes or bracketed descriptors, with an optional separate key
- Multilingual transcription in the original language, with optional translation into English or another target language
- Output formatted for NVivo, ATLAS.ti or MAXQDA import, or as Word, plain text or subtitle files
- Secure file transfer, confidentiality agreement on request and deletion after delivery
Not included
- Coding, summarizing or analyzing transcripts as part of this service; analysis support is available separately
- Guaranteeing identification of every speaker in large or noisy group recordings
- Recordings made without participant consent to recording, or files outside what your ethics approval permits us to handle
- Certified or sworn translation for legal purposes; see our certified document translation service
What you receive
- Transcripts
- One file per recording in your chosen format and style, with a header block for recording ID, date, duration and participants' codes.
- Transcription conventions sheet
- A one-page summary of the notation, style and de-identification rules used, suitable for your methods appendix.
- De-identification log (optional)
- A separate, securely delivered key linking codes to original identifiers, or confirmation that no key was retained, depending on your data management plan.
- Translations (optional)
- Parallel or separate translated transcripts, with translator notes on culturally specific terms that do not translate directly.
How it works
Agree the specification
We confirm verbatim style, speaker labels, timestamps, de-identification rules, languages, formats and deadlines, and receive your glossary.
Secure transfer
You upload recordings through a secure link or your institution's file-sharing platform. We confirm receipt and check audio quality.
Transcribe and review
Each recording is transcribed, then checked against the audio for accuracy, speaker attribution and consistent notation.
De-identify and format
Where agreed, identifiers are replaced consistently and the files are formatted for your analysis software.
Deliver and delete
Transcripts are delivered securely. Audio and working files are deleted on the agreed schedule, and we confirm deletion if requested.
Transcription styles and conventions
- Full (true) verbatim
- Every word, filler, false start, repetition and audible non-verbal event such as laughter or pauses. Preferred for conversation, discourse and some narrative analyses.
- Clean verbatim
- Removes fillers ('um', 'you know'), stutters and false starts while preserving the speaker's words and meaning. Common for thematic, content and framework analysis.
- Speaker labels
- Codes such as I, P01 or FG2-M3 applied consistently, matching the IDs in your participant tracking sheet.
- Timestamps
- At regular intervals (for example every two minutes), at speaker changes or at unclear passages, so you can return to the audio quickly.
- De-identification
- Direct identifiers replaced with codes or descriptors such as [colleague name] or [hospital in northern region], with indirect identifiers flagged for your review.
- Multilingual work
- Transcription in the source language by fluent transcribers, with optional translation handled with our academic translation team.
Verbatim or clean verbatim: which does your analysis need?
If your analysis looks at how something is said, such as hesitation around a sensitive topic, turn-taking in a focus group or the construction of identity in talk, choose full verbatim. If your analysis focuses on what participants experienced or believe, clean verbatim is usually enough and is easier to read and code.
Whichever you choose, clean verbatim should never 'correct' a participant's grammar, dialect or meaning. We remove only non-meaningful disfluencies and leave the participant's voice intact, which matters when you later quote them.
Confidentiality and de-identification in practice
Recordings often contain far more identifying information than researchers expect: colleagues' names, workplaces, towns, diagnoses. Your ethics approval and data management plan usually specify how this must be handled.
We apply your rules consistently across all files. Note that replacing names with codes while keeping a separate key is pseudonymization; under GDPR-style laws, pseudonymized data is still personal data. Truly anonymized transcripts need indirect identifiers to be considered too, and we flag passages where a combination of details could still identify someone so you can decide how to handle them.
- Files transferred through secure links rather than email attachments
- Access limited to the people working on your project
- Confidentiality agreements available and your institution's terms respected where required
- Deletion of audio and working copies on an agreed schedule
Multilingual recordings
When participants speak another language, or switch between languages, you may need a transcript in the original language for analysis and a translation for supervisors or publication. We agree whether code-switching is preserved, how culturally specific terms are handled, and whether translations should be close or idiomatic. For translating the instruments themselves, see research instrument translation.
What the work can look like
Illustrative example — fictional data
Twenty interviews on caregiving for an IPA study
A fictional doctoral researcher has 20 interviews, averaging 55 minutes, with family caregivers of people living with dementia. The planned analysis is interpretative phenomenological analysis, and the ethics approval requires de-identified transcripts with no retained key.
Specification
Clean verbatim was agreed, with pauses over three seconds and emotional expressions such as crying noted in brackets because they matter for interpretation. Timestamps were added every two minutes and at each inaudible passage.
De-identification
All names became role descriptors such as [husband] or [care home staff member], and towns became [town in region]. Eleven passages combining rare diagnoses with locations were flagged for the researcher to review.
Delivery
Transcripts were formatted as Word files with a consistent header for NVivo import, and audio files were deleted after the researcher confirmed receipt.
Illustrative, invented scenario. Your specification depends on your methodology, ethics approval and data management plan.
What you’ll need to provide
- Audio or video files, with a list of file names, durations and languages
- Your preferred transcription style and any conventions required by your supervisor or methodology
- A glossary of names, technical terms, acronyms and place names
- Your ethics approval conditions and data management plan requirements for handling recordings
- Speaker information (for example how many people and roles) to help with labeling
You don’t need everything ready to ask for a quote — tell us what you have and we’ll say what’s missing.
What affects price and turnaround
Price depends on
- Total audio duration and number of files
- Verbatim style and notation detail required
- Audio quality, number of speakers, accents and amount of crosstalk
- Languages involved and whether translation is needed
- De-identification depth and required output formats
Turnaround depends on
- Total recording hours and how quickly files can be released for work
- Audio quality and specialist terminology
- Availability of transcribers for less common languages
Every project is quoted individually. How we price our work.
Questions people ask before starting
Do you use AI transcription?
Automated speech recognition can be used for a first draft, always followed by a full human review against the recording. If your ethics approval or data management plan prohibits uploading recordings to third-party AI services, tell us and we will use a fully human process.
How accurate will the transcripts be?
Accuracy depends heavily on audio quality, accents, crosstalk and terminology. We review every file against the audio and mark uncertain passages with timestamps rather than guessing, so you can check them yourself.
Can you identify speakers in a focus group?
Usually, if participants introduce themselves and audio is clear. Where voices cannot be reliably distinguished, we label them as unidentified rather than risk wrong attribution. Video recordings or a seating plan help considerably.
Will you sign a confidentiality agreement?
Yes. We can sign a confidentiality agreement or work under your institution's data processing terms, and we can confirm deletion of files after delivery.
Is transcription something I am allowed to outsource for my thesis?
Many institutions allow professional transcription, but some expect students to transcribe their own data as part of familiarization with it. Check your program's rules and your ethics approval, which may need to name a transcription service.
What formats can you deliver?
Word, plain text, PDF, or files structured for NVivo, ATLAS.ti or MAXQDA import. Subtitle formats are available for video.