Medical Transcription Service UK 2026-2027

Transcription looks like typing. It is actually the first analytical decision you make about your data, and the wrong convention can quietly make your intended analysis impossible.

Projectsdeal provides academic and research transcription for health and medical researchers, dissertation students and doctoral candidates: qualitative interviews, focus groups, oral history recordings and lectures, transcribed in the convention your analysis actually requires. We work to verbatim, intelligent verbatim or denaturalised conventions, apply a notation key you can reproduce in an appendix, and handle clinical terminology accurately rather than phonetically. All work is human, produced under our Zero AI Policy, and handled under UK GDPR with written confidentiality terms. We do not transcribe live patient records or anything constituting a clinical health record.

Medical Transcription Service for Research Audio, Interviews and Academic Work

Transcription is the least glamorous stage of qualitative research and the one that quietly decides how good the rest of it can be. Researchers who spent months on ethics approval, recruitment and careful interviewing often treat the resulting audio as an administrative obstacle standing between them and the analysis. It is not. The transcript is the object you will actually analyse, and whatever it does not record is, for practical purposes, gone. Our Medical Transcription Service exists for researchers, dissertation students and doctoral candidates whose health and medical research audio deserves the seriousness they gave to the interviewing itself.

Projectsdeal has supported UK students and researchers since 2001, across more than 115,000 orders at an average rating of 4.9/5, with 120+ PhD-qualified writers and researchers. Transcription is matched to subject knowledge rather than allocated to whoever is free, because health research audio is full of terminology a general transcriber will render phonetically and therefore wrongly. Everything is produced by people under our Zero AI Policy, handled under UK GDPR with written confidentiality terms available, and delivered in the convention your analysis actually requires. Ordering runs online 24x7, with WhatsApp support on +447447882377.

One boundary belongs at the top of this page rather than in the small print. We do not transcribe live patient records, dictated clinical notes, clinic letters, operating notes or anything that would form part of a clinical health record. That is regulated clinical documentation sitting inside governance arrangements we are not part of, and it is not what an academic support company should be touching. Our work is academic and research transcription. Any audio containing identifiable patient information must have the appropriate ethics and information governance approvals in place before it is shared with anybody at all, including us, and if material appears to fall outside them we will say so rather than proceed.

Transcription Is a Methodological Decision, Not a Clerical Task

The single most useful idea on this page is that the transcription convention you choose determines what analysis is possible afterwards. A transcript is not a neutral record of an event. It is a representation, made by somebody, according to rules, and those rules decide what survives from the room into the text. If the rules delete filler, hesitation and overlap, then no amount of later cleverness can analyse hesitation, because the evidence of it no longer exists. Researchers who discover this after transcription is finished are left either re-transcribing the corpus or changing the analysis they promised in the proposal.

Verbatim, Intelligent Verbatim and Denaturalised Transcription

Three broad conventions cover most academic work, and the vocabulary is used loosely enough to be worth stating precisely. Full verbatim records everything audible: repeated words, false starts, fillers, breaths, laughter and pauses. Intelligent verbatim keeps the speaker's own words and meaning but removes disfluency and lightly repairs sentence structure. Denaturalised transcription goes further towards readability, tidying grammar and removing the idiosyncrasies of speech to foreground content. Against these sits the naturalised end of the spectrum, where conversation analysis lives and where micro-pauses, overlap, intonation and in-breaths are recorded because in that tradition they are the data.

The error we see most often is the mismatch in the middle two rows. A student plans a thematic analysis, orders full verbatim because it sounds rigorous, then wades for weeks through text cluttered with material they will never code. The opposite error is worse and far less recoverable: a student plans something interactional, orders clean intelligent verbatim because it is quicker and cheaper, and arrives at analysis to find the phenomenon they intended to study has been edited out of the evidence. The first mistake costs time. The second can cost the study.

Notation Conventions and Why You Need a Key in Your Appendix

Once a transcript records anything beyond plain words it needs a notation system, and once it has one the reader needs the key. The best-known scheme is the one developed by Gail Jefferson for conversation analysis. It is not the only option, and simpler in-house schemes are respectable for interview work provided they are applied consistently and documented. What is not respectable is a transcript peppered with brackets, dots and arrows that an examiner must decode by inference. Include the key as an appendix and refer to it from the methodology; two lines and one page convert typographical marks into evidence of methodological control.

Laughter, Hesitation and False Starts: The Analytic Value of Disfluency

Speech is not written prose delivered aloud. People start sentences and abandon them, correct themselves mid-clause, repeat words while thinking, laugh at things that are not funny, and go quiet at the point where a question has landed hardest. Written convention treats all of that as error; qualitative research frequently treats it as the most informative material in the interview, because it is where the speaker is visibly managing something rather than reporting it. Consider a clinician asked about a decision that went badly: the words may be measured, while the three-second silence, the laugh and the two abandoned sentence openings say something else entirely.

Focus Groups, Multiple Speakers and the Problem of Attribution

Focus group transcription is substantially harder than one-to-one transcription and is routinely underestimated when researchers plan time and budget. In a one-to-one interview there are two voices and turn-taking is usually orderly. In a group of eight there are similar voices, side conversations, people agreeing over the top of each other, and moments where three speak at once and none is fully recoverable. Attribution rather than audibility is the core difficulty, and it damages analysis quietly rather than announcing itself.

Attribution matters because most focus group analysis wants to know who said what. If you intend to compare junior and senior staff, or patients and carers, a contribution attributed to the wrong person makes the analysis wrong in a way nobody can later detect. We therefore mark contributions we cannot confidently attribute as an unidentified speaker rather than assigning them to the likeliest candidate. You can make attribution far easier at the recording stage: ask participants to state a pseudonym or number before their first few contributions, note the seating order, have the moderator name people when inviting them to speak, and use two devices at different points around the table.

Accent, Dialect and Code-Switching: An Ethical Question, Not a Typing One

How you render a participant's speech on the page is an ethical decision with a long and uncomfortable history in social research. Transcribe dialect phonetically and you produce a text in which some participants appear articulate and others do not, purely as a function of how close their speech is to standard written English. Readers, including examiners and reviewers, respond to that difference whether or not they intend to. Standardise everything instead and you erase features of speech that may be central to identity, community membership and the meaning of what was said. There is no universally correct answer, which is why the decision must be made deliberately and then reported. Our default is a middle position: preserve the speaker's own vocabulary, grammar and word order, but use standard spelling rather than eye dialect, so distinctive constructions survive while speech is not rendered as a phonetic curiosity. Where a study is specifically about language, community or migration, we apply a more faithful rendering, because there the linguistic detail is the evidence.

Code-switching raises the same question more sharply. Participants who move between languages, or between a community variety and standard English, are usually doing something meaningful at the switch: marking intimacy, distance, register or the exclusion of an outsider, and flattening it into one language destroys the phenomenon. We mark the switch, transcribe the passage in the language used, and supply a bracketed translation so a monolingual reader can follow without the original being erased. Studies in this territory usually need a careful research ethics framing too, since representation of participants is itself an ethical matter.

Translated Interviews and Why the Translation Decision Must Be Reported

Translation is not a technical conversion; it is interpretation performed by somebody with their own linguistic and cultural position. Hedging, politeness forms, idiom, kinship terms and illness concepts frequently have no clean English equivalent, and the translator's choice among imperfect options shapes what your analysis can subsequently see. A participant who used a term for a bodily complaint carrying social meaning in their community has not said the same thing as the nearest English clinical label, and translating it as that label silently deletes the finding you were looking for.

Report the decisions rather than hiding them. Your methodology should state whether transcription happened in the source language before translation or directly into English as it was transcribed, who translated and what their relationship to the research was, whether back-translation occurred, and how contested terms were resolved. Where budget allows, retaining the original-language transcript alongside the English version is good practice, since it permits verification later and lets quotations be presented in both forms in the thesis.

Timestamps, File Naming and Working Between Audio and Text

Timestamps are the cheapest quality improvement available in research transcription, and researchers who have worked without them rarely go back. They let you return to the exact moment a quotation occurred, which you will want more often than you expect: when a passage reads ambiguously, when a supervisor questions an interpretation, when you are choosing quotations and want to hear the tone first, and when a reading is challenged in a viva or in peer review.

Audio Quality: What Can Be Improved and What Cannot Be Recovered

Some audio problems can be reduced and others are permanent. Constant hiss, mains hum, mild room reverberation and uneven levels between a loud interviewer and a quiet participant can usually be improved enough to work with, and an occasional door or a single overlapping exchange can be handled by replaying the passage slowly. What cannot be recovered is speech that was never adequately captured. A device left at the far end of a table in a room with hard surfaces, two people speaking simultaneously into one low-quality microphone, sustained loud background noise, or a participant who drops to a whisper at the emotionally significant moments will all produce passages that simply are not there. Nothing reconstructs missing audio, and anybody claiming otherwise is producing plausible guesses, which in health research is worse than an honest gap. The remedies are cheap and all at the recording stage: an external or lapel microphone, the quietest room available, the device placed between speakers, and a test playback of the first minute before you continue.

Turnaround: The Ratio of Audio Hours to Working Hours

The most common planning error in qualitative projects is treating transcription time as roughly equal to audio time. It never is. The ratio depends on the number of speakers, the clarity of the recording, the density of specialist terminology and, above all, the level of notation required. The table gives realistic working ratios rather than optimistic ones, and they apply just as much to a researcher transcribing their own data as to a professional doing it.

Transcribe It Yourself or Outsource? The Familiarisation Argument

There is a serious methodological case for doing your own transcription, and it deserves stating properly rather than dismissing as purism. Transcribing an interview means listening several times, at slow speed, with full attention on the words. That is immersion, and it is why many qualitative methodologists treat transcription as the first stage of analysis rather than a precondition for it. Researchers who transcribe their own data usually arrive at coding already knowing where the interesting material sits, which contradictions recur, and which questions produced discomfort.

What you lose by outsourcing is that immersion, not accuracy. The usual objection to transcription services is framed as a quality concern when the real cost is analytic familiarity: a subject-matched transcriber will generally produce a more accurate text than a tired researcher working at two in the morning, but cannot hand over tacit knowledge of the corpus. The workable compromise, which we recommend to doctoral candidates, is to transcribe a purposive subset yourself, outsource the rest, then listen to every outsourced recording while reading its transcript and annotating as you go. That recovers most of the familiarisation, doubles as a quality check, and can be described honestly in the methodology. The chapters that follow are supported by our literature review writing services and thesis editing service.

Medical Terminology, Abbreviations and the Drug-Name Problem

This is where a general transcription service and a subject-matched one visibly diverge. A transcriber without clinical or scientific background writes what a term sounds like. In ordinary speech that yields a mildly odd word a reader corrects mentally; in health research it yields a word that is wrong in a way that changes meaning, and often one that looks plausible to anybody outside the field, so it survives proofreading and reaches the examiner intact. Drug names are the sharpest version: pharmaceutical nomenclature contains many names close in sound while belonging to different classes with different indications and risks, so a guess between two of them changes what the participant said about clinical practice.

Abbreviations carry a second-order risk, because the same letters mean different things in different specialties and participants use them constantly without expansion, everybody in their own working environment understanding them. A responsible transcript preserves the abbreviation as spoken rather than expanding it into an assumed meaning, leaving interpretation to the researcher who knows the context. Where a term or a dose is unclear it is flagged with a timestamp and a bracketed best guess, never silently resolved, and a project glossary supplied in advance reduces ambiguity sharply. Researchers writing up often pair transcription with our medicine essay writing service or nursing essay writing service.

Confidentiality and Data Protection: Audio Is Personal Data

A recording of an identifiable person speaking is personal data in its own right, because a voice identifies the speaker independently of anything said. Where the content concerns physical or mental health, as health research audio routinely does, it is also special category data under UK GDPR, attracting additional protection and requiring an identified condition for processing on top of a lawful basis. Researchers sometimes treat audio as a working file and the transcript as the real data. The law does not see it that way, and neither should your data management plan.

In practice the recording needs the same care as the transcript from the moment it exists: secure transfer rather than an email attachment, encrypted storage rather than a desktop folder, named access rather than an open shared drive, and a defined deletion point for audio and transcript alike. We work inside that framework rather than around it. Files move by secure link, are held only for as long as the work requires, are seen only by the assigned transcriber and the order coordinator, and are deleted on your confirmation, which we put in writing. We will also sign your institution’s confidentiality or data processing agreement before any file is sent.

Anonymisation and Pseudonymisation Are Not the Same Thing

These two words are used interchangeably in student writing and mean materially different things. Pseudonymisation replaces direct identifiers with codes or invented names while a key linking code to person still exists somewhere. Anonymisation means the data can no longer be attributed to an identifiable individual by any means reasonably likely to be used, and no key exists anywhere. The legal consequence is significant: pseudonymised data remains personal data and stays fully within the scope of data protection law, whereas genuinely anonymous data falls outside it.

Most transcripts described in methodology chapters as anonymised are in fact pseudonymised, because the researcher retains a participant list in order to honour withdrawal requests. That is legitimate and usually the right design; describing it with the more reassuring word is not. Effective de-identification also goes well beyond removing names. Job titles in small specialties, ward names, unusual conditions and the combination of role plus location can each identify a person to a reader working in the same system, and in health research that is exactly who reads your thesis. We replace identifiers with bracketed descriptors that preserve the analytic category and flag passages where surrounding detail may still identify, so the judgement is yours rather than a surprise at examination.

Ethics Committee Requirements and Confidentiality Agreements

Research ethics committees want to know who will have access to identifiable data, and an external transcriber is precisely such an access point. It should therefore appear in the ethics application, in the participant information sheet and, where the committee requires it, in the consent form. Committees commonly ask for confirmation that a confidentiality agreement is in place with the transcriber, a description of how files will be transferred and stored, and a stated retention and deletion schedule for both audio and transcript.

The failure mode is avoidable and unfortunately common: a participant information sheet promises that only the research team will hear the recording, and the researcher, later drowning in audio, sends it to a transcription service anyway. That is an ethical breach however securely the service handles the file, because consent was given on the basis of a statement that has been broken. Include the possibility at application stage even if you are unsure you will use it, since an amendment after recruitment is far more work than a sentence written in advance. Our research ethics assignment help page covers how these applications are built and assessed.

Transcription in the Dissertation Methodology Chapter

Examiners read methodology chapters looking for evidence that decisions were made rather than defaults accepted, and transcription is a short subsection offering a cheap opportunity to demonstrate exactly that. A strong treatment names the convention, justifies it against the analytic approach, explains how non-verbal features and identifiers were handled, states who transcribed and under what agreement, and points to the notation key in the appendices.

The weak version is one sentence saying interviews were transcribed verbatim, which raises more questions than it answers, since under questioning they rarely turn out to have been verbatim in any technical sense. Vivas probe this because it is a quick test of whether a candidate understands their own data as constructed, and being able to say why you removed filler and kept laughter is a strong moment in an oral examination. If you are drafting that chapter, our dissertation writing support shows the expected shape, and our proofreading services handle the final consistency pass across extracts, participant codes and appendix cross-references, where late errors cluster.

Oral History, Lectures and Accessibility Transcription

Not all academic transcription is interview data. Oral history has its own conventions: recordings are often long, the archival value may outlive the project, names and places need verifying rather than guessing, and the ethical relationship with the speaker frequently includes a right to review the transcript. We treat oral history work as archival rather than disposable, which means careful handling of proper nouns and a strong preference for preserving the speaker's own voice rather than tidying it into standard prose.

Lecture and seminar transcription serves a different purpose. Students with specific learning differences or hearing impairments, and those studying in a second language, often need a text record to work from, and university disability support arrangements frequently recognise that need. The technical challenge differs too: distant microphones, questions from the floor never audible to the recorder, and references to slides the audio does not describe. We mark unrecoverable audience contributions honestly and note where the speaker referred to visual material, so the transcript does not silently pretend to be complete.

Common Transcription Mistakes and How to Avoid Them

Almost every problem we are asked to fix falls into a small number of categories, and nearly all are cheaper to prevent than to remedy. The table is worth reading before your first recording rather than after your tenth.

How Our Transcription Process Works

Formats suit what you will do next. Word documents with speaker labels and timestamps suit reading, quoting and supervision; plain text and rich text exports suit import into the qualitative analysis software commonly used in UK doctoral work. Where a project has a quantitative strand as well, we align participant identifiers so the two datasets can be matched, which our data analysis team can then work with directly.

What We Will and Will Not Do

We transcribe research and academic audio: qualitative interviews, focus groups, oral history recordings, lectures and seminars, viva rehearsals, conference presentations and research meetings where you hold the necessary approvals. We produce transcripts to your stated convention, apply notation consistently, de-identify to your rules, supply timestamps and a notation key, and return the material in the format your analysis needs.

We do not transcribe live patient records, dictated clinical notes, discharge summaries, clinic letters, operating notes or any material forming part of a clinical health record. We do not accept audio containing identifiable patient information unless the appropriate ethics and information governance approvals are in place and evidenced, and we decline material that appears to fall outside them. We do not fabricate content, we do not fill inaudible gaps with plausible words, and we do not transcribe recordings made without the consent of the people recorded. On the writing side the usual boundary applies: Projectsdeal supplies bespoke model answers and reference material written to your brief, to be studied and learned from rather than submitted, supported where useful by our essay writing service, assignment help and UK essay writers teams.

Why Researchers Choose Projectsdeal for Academic Transcription

Practical terms are the same across everything we do: on-time delivery, money-back protection, free unlimited revisions within the brief, GDPR-compliant confidentiality, and instalments on larger orders. We do not contact your institution, and your audio, transcripts and briefs are never reused, resold or shown to anybody outside the team working on your order. If you are still planning, talk to us before you record: ten minutes on the convention, the consent wording and the recording setup routinely saves weeks later. Support runs from the research proposal through the literature review to the final dissertation.

How It Works — 3 Steps, Open 24x7

Join 115,000+ UK students since 2001 • ✅ Zero AI • ✅ No hidden fees • ✅ Money-back guarantee

Zero AI Policy — Proven on Every Order

UK universities scan submissions with AI detectors, and flagged work triggers misconduct panels. Our Zero AI Policy is absolute: no AI writes any part of your work, ever. Every order is written by a named human academic with a UK degree in your subject, then verified through Turnitin’s AI and similarity checkers — and both reports are yours free, so you hold independent proof of 0% AI and 0% plagiarism before you submit. That protection comes standard with every medical transcription service order.

Our Guarantees, In Writing

What UK Students Say

Frequently Asked Questions

1. What is the difference between verbatim and intelligent verbatim transcription?
Verbatim transcription records everything that was said exactly as it was said, including false starts, repetitions, filler words, stammers and audible non-verbal sounds such as laughter or sighing. Intelligent verbatim removes those features and lightly tidies the speech into readable sentences while keeping the speaker's own words and meaning intact. Verbatim preserves more analytic material and takes considerably longer to produce; intelligent verbatim is faster to read and far easier to code thematically. The decision should follow your analytic method rather than your preference, and whichever you choose should be stated in your methodology chapter so a reader knows what the transcript does and does not represent.

2. Which transcription style should I use for thematic analysis?
Most thematic analysis is carried out well from clean intelligent verbatim, because the unit of interest is what participants say about a topic rather than the fine mechanics of how they say it. Removing filler and false starts makes codes easier to see and reduces the volume of text you have to work through repeatedly. That said, you should keep laughter, long pauses and clear emotional markers, because those often signal discomfort, irony or hesitation that changes how a statement should be read. If your analysis has any discursive or interactional component, move towards full verbatim instead. Say in the methodology which convention you used and why.

3. Do I need Jefferson notation for my interviews?
Only if you are doing conversation analysis or a closely related interactional method. Jefferson notation was developed for conversation analysis and captures overlap, latching, in-breaths, stretched sounds, volume, pitch movement and precisely timed pauses, all of which are analytic data in that tradition. Applying it to a straightforward thematic study creates a transcript that is slow to produce, hard to read and full of detail you will never use. Conversely, running conversation analysis on an intelligent verbatim transcript is not possible, because the features you need to analyse have already been deleted. Decide before transcription begins, not after.

4. How long does it take to transcribe one hour of audio?
Far longer than an hour, and the ratio depends on the recording rather than on typing speed. Clean one-to-one audio with a single clear speaker sits at the fast end; a multi-speaker focus group with overlapping talk, background noise and strong regional accents sits at the slow end; full conversation-analytic notation is slower again by a wide margin. Clinical terminology adds time because terms must be checked rather than guessed. If you are transcribing your own data, budget generously and treat the time as analysis rather than admin, because that is what it becomes.

5. Should I transcribe my own research interviews or outsource them?
There is a genuine analytic argument for doing it yourself: transcription is a stage of familiarisation, and researchers who transcribe their own interviews usually arrive at coding already knowing where the interesting material is. What you lose by outsourcing is that immersion, not accuracy. The practical counter-argument is time, and a researcher with forty hours of audio and a submission date is not making a methodological choice so much as an arithmetic one. A workable compromise is to transcribe a subset yourself, outsource the rest, and then listen to every outsourced recording while reading its transcript so the familiarisation still happens.

6. Do you transcribe patient records or clinical notes?
No. We do not transcribe live patient records, clinic letters, dictated notes, operating notes or anything that would form part of a clinical health record. That is regulated clinical documentation and it belongs with providers operating inside the relevant clinical governance arrangements, not with an academic support company. Our work is academic and research transcription: research interviews, focus groups, oral history, lectures and seminars. If audio you hold contains identifiable patient information, it must have the appropriate ethics and information governance approvals in place before it is shared with anybody, including us.

7. Is interview audio classed as personal data under UK GDPR?
Yes, and usually more than that. A voice recording of an identifiable person is personal data in itself, because the voice can identify the speaker, and the content of a health research interview will frequently include information about physical or mental health, which is special category data attracting additional protection. That is why research recordings need a lawful basis, a documented condition for processing special category data, secure storage, controlled transfer and a defined retention and deletion point. Treating audio as though it were less sensitive than the transcript is a common and serious mistake.

8. What is the difference between anonymisation and pseudonymisation?
Pseudonymisation replaces direct identifiers with codes or pseudonyms while a key that allows re-identification still exists somewhere. Anonymisation means the data can no longer be attributed to an identifiable individual by any reasonably likely means, and no key exists. The distinction matters legally, because pseudonymised data remains personal data and stays within the scope of data protection law, whereas genuinely anonymous data does not. Most research transcripts described as anonymised are in fact pseudonymised, because the researcher retains a participant list. Describe what you have actually done rather than using the more reassuring word.

9. How do you handle strong accents and dialect in transcription?
Carefully, and with the ethical question made explicit rather than resolved silently. Rendering dialect phonetically can make a participant look uneducated on the page in a way they did not sound in the room, and that misrepresents them. Standardising everything into neutral written English removes features that may be central to identity, community and meaning, particularly in research about culture, migration or marginalised groups. Our default is to preserve the speaker's own vocabulary, grammar and word order while using standard spelling, and to flag anything genuinely ambiguous rather than smoothing it away. If your study needs a different approach, tell us and we will apply it consistently.

10. Can you transcribe focus groups with several speakers?
Yes, though focus groups are substantially harder than one-to-one interviews and should be planned for accordingly. Overlapping speech, participants who sound similar, people talking across each other and side conversations all reduce what can be attributed with confidence. The single most useful thing a researcher can do is have participants say their pseudonym or number at the start and, ideally, before their first few contributions, plus a moderator note of the seating order. Where a contribution genuinely cannot be attributed we mark it as an unidentified speaker rather than guessing, because a wrong attribution corrupts any analysis that compares participants.

11. How accurate is transcription of medical terminology?
Accuracy depends almost entirely on whether the person transcribing understands the field, which is why we match transcribers to subject area. A transcriber without clinical or scientific background writes what a term sounds like, and in medicine that produces errors that change meaning rather than merely looking untidy. Drug names are the sharpest risk, because many are similar in sound while being entirely different in class, dose and use. Abbreviations are the second risk, since the same letters carry different meanings in different specialties. We check terms against reference sources and flag genuinely unclear audio rather than inventing a plausible word.

12. Do you provide timestamps in transcripts?
Yes, and we recommend them for research work. Timestamps let you return to the audio at the exact point a quotation occurs, which matters when you are checking tone, verifying a contested passage, or selecting quotations for a results chapter and want to hear them again before committing. They are also useful during supervision, because a supervisor can be pointed to a moment rather than a page. You can specify the interval, whether at fixed intervals, at each speaker change, or only at points you mark. Speaker-change timestamps are usually the most useful setting for interview data.

13. What audio quality do you need, and can poor recordings be fixed?
Some problems can be reduced and others cannot be recovered at all. Background hiss, mild room echo and uneven volume can often be improved enough to work with. Speech that was never captured because the microphone was too far away, two people talking simultaneously into a single low-quality device, or a recording ruined by sustained loud background noise cannot be reconstructed, and anybody promising otherwise is guessing at words. The fixes are all at the recording stage: use an external microphone where possible, record in a quiet room, place the device between speakers, and check the first minute back before continuing.

14. Can you transcribe interviews conducted in another language?
We can arrange transcription and translation for common language pairs, and the important point is methodological rather than practical. Translation is interpretation, and the decisions made in translating idiom, hedging and culturally specific terms shape what your analysis can see. You should record in your methodology whether you transcribed in the original language and then translated, or translated directly while transcribing, who did the translation and what their relationship to the research was, and how any disputed terms were resolved. Where budget allows, keeping the original-language transcript alongside the translation is good practice and allows verification later.

15. How do you keep my research audio confidential?
Transfers use secure links rather than ordinary email attachments, files are held only for as long as the work requires, and access is limited to the transcriber assigned and the coordinator managing the order. We will sign a confidentiality agreement where your ethics approval or your institution requires one, which is increasingly standard and entirely reasonable. Files are deleted on confirmation that the transcript has been received and accepted, and we can confirm deletion in writing for your records. We do not contact your institution, and your material is never reused, resold or shown to anybody else.

16. Does using a transcription service need to go in my ethics application?
Almost always, yes. Research ethics committees generally want to know who will have access to identifiable data, so if a third party will hear the recordings, that should be stated in the application, in the participant information sheet and, where relevant, in the consent form. Committees commonly ask for a confidentiality agreement with the transcriber and for a description of how files will be transferred, stored and deleted. Telling participants that only the research team will hear the audio and then sending it to an external transcriber is a genuine ethical breach, so get the wording right at application stage rather than afterwards.