Recording Research Participants: Consent, Ethics and Audit Trails

Reviewed and fact-checked: 21 July 2026.

Research audio can preserve language, emphasis and context that field notes miss. It also creates identifiable data that may expose participants, communities and third parties long after the interview ends. AI transcription adds suppliers, derived outputs and potential reuse that must be designed into the protocol.

This guide covers practical UK research-ethics and data-governance questions. It is general information, not a substitute for institutional ethics review, sponsor requirements or legal advice.

Design recording into the protocol

Document:

• Why audio is necessary
• Which participants and settings are involved
• Whether remote, phone or in-person capture is used
• How consent will be obtained and recorded
• Which AI supplier performs transcription
• How identifiers and third-party data are handled
• Who can access raw and processed data
• How quotations, archives and future reuse work
• Retention and destruction dates

A field researcher should not switch to a convenient consumer app without an approved amendment.

Obtain the required ethics and governance approvals

The Health Research Authority and Medical Research Council provide current guidance on participant information and consent, stressing that forms should be appropriate and proportionate rather than copied mechanically from a template.

Official source: HRA guidance on informing participants and seeking consent.

UKRI ethics guidance expects protocols to address freely given and adequately informed consent, risk, secure data handling, sharing, confidentiality, privacy and data management.

Official source: UKRI ethics-review guidance.

Explain the complete data journey

Participants should understand:

• That audio will be recorded
• The purpose
• Whether AI transcription or summaries are used
• Where processing occurs
• Who can hear audio and read text
• Whether identifiers are removed
• How quotations will be attributed
• Whether data may be archived or reused
• Realistic withdrawal limits
• When raw and processed files will be destroyed

Do not bundle optional permissions

Participation, audio recording, direct quotation, identifiable publication, archive deposit and future research reuse may need separate choices. A participant may agree to the interview and transcription but decline publication of their voice.

Choose an appropriate consent method

Written consent is not universally required. UKRI notes that verbal consent can be ethically preferable where signatures create risk, but the approach and justification should be included in ethics review. Whatever method is used, record what information was provided and what the participant agreed to.

Official source: UKRI research-consent guidance.

Make withdrawal limits honest

State the point after which withdrawal may be impossible—for example after anonymised aggregation, publication or secure archive deposit. Do not promise deletion “at any time” when copies may already be incorporated into analysis or outputs.

Assess sensitive and vulnerable-participant risk

Research may reveal health, political opinions, ethnicity, sexuality, criminal activity, migration status or trauma. Identify the Article 6 lawful basis, Article 9 condition and safeguarding or disclosure procedure. Avoid collecting sensitive information that the research question does not require.

Approve the transcription supplier

• Controller and processor roles
• Confidentiality and contract terms
• Subprocessors and data locations
• International-transfer safeguards
• Whether content trains models
• Security and authentication
• Deletion and backup handling
• Support for participant rights
• Account closure and data return

Do not upload identifiable recordings into a personal AI account.

Use fieldwork safeguards

• Confirm consent before recording.
• Use a neutral file reference rather than participant name.
• Keep the device under researcher control.
• Avoid recording bystanders.
• Pause if a participant becomes distressed.
• Use the approved response for disclosures or danger.
• Transfer data promptly through the secure route.
• Delete device copies under the protocol.

Correct transcripts without rewriting participants

Verify names, numbers, key quotations and specialist terms. Preserve uncertainty, dialect and meaning where relevant. Maintain a clear distinction between:

• Verbatim speech
• Researcher clarification
• Translation
• AI-generated summary
• Analytical coding

Keep analysis auditable

Record transcript corrections, coding decisions, AI prompts, generated themes, researcher overrides and version history where methodological integrity requires it. AI-generated themes should be tested against source material and alternative interpretations.

Pseudonymisation is not anonymity

Removing names does not automatically anonymise rare experiences, workplaces, villages or quotations. Assess re-identification from context and public information. Keep any re-identification key separately with tighter access.

Plan quotations and publication

Decide whether quotations are attributed, pseudonymised, translated or paraphrased. Searchable distinctive phrases can reveal identity. For public audio clips, voice itself may identify the speaker.

Set a research data-management plan

Assign separate rules to:

• Raw audio
• Master transcript
• Redacted analysis copy
• Consent evidence
• Codebook and audit trail
• Published outputs
• Archive deposit
• Backups and exports

Using NERALVO Halo for approved fieldwork

NERALVO Halo provides NOTE mode, CALL mode for supported and permitted calls, 64GB local storage, up to 35 hours of recording and Bluetooth connection to DOWAY for transcripts, summaries, templates, translations, mind maps and exports. One year of DOWAY Max is included.

Researchers should accurately describe local capture and DOWAY processing in the protocol and participant materials. The device does not provide ethics approval or automatically anonymise the data.

Research-recording checklist

• Audio necessity documented
• Protocol and ethics approval obtained
• Participant information matches reality
• Recording, quotation and reuse choices separated
• Consent method and evidence approved
• Withdrawal limits stated honestly
• Sensitive-data and safeguarding risks assessed
• Supplier and transfers approved
• Fieldwork security planned
• Transcript corrections and analysis auditable
• Re-identification risk tested
• Retention, archive and destruction dates assigned

Frequently asked questions

Is verbal consent valid?

It can be appropriate in some research, but follow the approved protocol and document the justification and evidence.

Can a pseudonymised transcript be shared openly?

Not automatically. It may still be personal data and identifiable through context.

Can old recordings be used for a new study?

Only after assessing the original permissions, ethics approval, lawful basis, participant expectations and archive terms.

Can AI generate the final themes?

It can assist analysis, but researchers must test outputs, document decisions and remain accountable for interpretation.

Respect participants beyond the interview

Ethical recording covers the complete lifecycle: informed choice, safe capture, controlled AI processing, faithful interpretation, responsible quotation and the promised end of the data.