Should Identifiable Recordings Be Kept After Transcription Is Complete?

Once you have the transcript, do you still need the original recording?

You have finished interviewing participants. The audio or video has been transcribed, the transcript has been checked, and perhaps identifying details have already been removed from the working text.

Now there is an obvious question: should you delete the original recordings?

There is no universal rule that says identifiable recordings must be destroyed immediately after transcription, just as there is no universal rule requiring researchers to keep them. Transcription is an important milestone in the data lifecycle, but it does not by itself determine the appropriate retention period.

Keep the recording only when there is a continuing reason to keep it

Transcription and retention answer two different questions

A transcript does not necessarily replace everything in the recording

Transcription converts recorded material into text, but audio and video may contain information that the transcript does not fully preserve. Tone, pauses, emphasis, overlapping speech, pronunciation, emotional expression, and other vocal characteristics may matter in some forms of analysis. Video can contain gestures, facial expressions, interactions, spatial relationships, or other visual information that disappears from a conventional transcript.

For a study concerned only with the semantic content of interview responses, those features may have little analytical value after the transcript has been verified. For discourse analysis, conversation analysis, multimodal analysis, or other approaches in which speech or visual behaviour forms part of the evidence, the original recording may remain central research data.

The first question is therefore not simply whether transcription is complete. Ask whether the original recording still contains information necessary for the research purpose.

Recordings may be useful for checking transcription accuracy

A transcript can contain errors. Words may be misheard, technical terminology may be transcribed incorrectly, speakers may be confused, and automated transcription can introduce substitutions that alter meaning.

Keeping the original recording for an appropriate period can allow researchers to verify uncertain passages and audit the accuracy of the transcript. If this is the reason for retention, however, the period should correspond to that purpose. "We may need to check the transcript" is different from "we will keep every recording indefinitely."

Retention may support re-analysis or research verification

Some projects legitimately require researchers to return to the original material during coding, interpretation, quality assurance, collaborative analysis, or preparation of publications. In longitudinal research, recordings may also form part of a planned dataset that will be revisited over time.

Whether such retention is appropriate depends on the study. Researchers should be able to explain why access to the original identifiable material remains necessary rather than assuming that retaining more data is always scientifically preferable.

That distinction matters because identifiable recordings preserve information that may be difficult to remove. A participant's voice can remain identifying even when no name is spoken, while video may preserve faces, surroundings, voices, and other characteristics.

Data protection rules may require you to justify retention

Retention requirements differ among jurisdictions, institutions, and research settings. Researchers should therefore avoid treating any single retention period as a universal standard.

Under the UK GDPR storage-limitation principle, for example, personal data generally should not be kept in identifiable form for longer than necessary for the purposes for which they are processed. Current Information Commissioner's Office guidance says organizations should be able to justify retention periods, review whether personal data are still needed, and erase or anonymise data when they no longer need them.

Research receives an important qualification under that framework. Personal data may be retained for longer, including for scientific or historical research or public-interest archiving, when the relevant requirements and safeguards are satisfied. This means that "storage limitation" should not be simplified into "delete everything as soon as possible."

Required research records are not necessarily the same as raw recordings

This distinction prevents another common mistake. A rule requiring certain research records to be retained for a specified period does not automatically mean that every raw audio or video file must be kept for exactly the same period.

For example, HHS regulations require institutions to retain certain IRB records for at least three years after completion of research, and OHRP explains that investigators designated to maintain informed-consent documentation may have corresponding record-retention responsibilities. OHRP also explicitly notes that other regulations or policies may apply to study-data retention.

Researchers should therefore identify what a retention requirement actually covers. Ethics records, signed consent forms, coded datasets, transcripts, linkage files, and raw recordings can have different retention requirements.

What participants were told matters

If participants were told that recordings would be destroyed after transcription, after verification, at the end of the project, or after a specified period, that commitment should not casually be replaced with indefinite retention.

Conversely, participants may have agreed to longer-term preservation or future research use under an approved arrangement. The correct action depends on the consent process, applicable governance requirements, and what was actually approved.

Informed-consent requirements under the Common Rule include describing the extent to which confidentiality of identifying records will be maintained. More broadly, researchers should make material data-handling practices sufficiently clear that participants are not surprised by what happens to identifiable recordings later.

Withdrawal does not always require destruction of already collected recordings

Researchers should also avoid promising participants an unconditional right to have every piece of previously collected research data destroyed unless that is actually the applicable policy.

For HHS-conducted or supported research, OHRP guidance states that when a participant withdraws from an ongoing study, an investigator may retain and analyze already collected data, including identifiable private information. Other regulatory systems, consent arrangements, institutional policies, and study designs may produce different obligations.

The practical lesson is to explain withdrawal and data-retention arrangements accurately rather than improvising them after a participant asks.

Keeping recordings creates continuing confidentiality obligations

Retention is not passive. As long as an identifiable recording exists, it must continue to be managed appropriately.

That means considering storage security, access controls, backups, copies, transfers, personnel changes, third-party services, and eventual disposal. The longer identifiable material is retained, the longer those governance responsibilities continue.

This is one reason to distinguish the original recording from a less identifying working dataset. If analysis can proceed using an appropriately de-identified transcript, access to the original may sometimes be restricted to a smaller group even while legitimate retention continues.

The broader principles for handling identifiable audio, video, and photographs continue to apply for as long as those originals are retained.

Archiving can be legitimate, but it should be planned

Some recordings have genuine value as research records or reusable qualitative data. Long-term preservation may therefore be appropriate, particularly when future analysis requires information that cannot be represented adequately in a transcript.

Archiving identifiable recordings should nevertheless be intentional. Researchers need to consider participant consent, ethics approval, repository requirements, access conditions, applicable law, and whether open, safeguarded, or controlled access is appropriate.

Long-term preservation is not the same thing as leaving forgotten recordings in a researcher's personal storage account. Academic archaeology is charming only when the artifacts are supposed to be there.

Deciding what to do after interview transcription

The important point is that transcription triggers a retention review. It does not mechanically dictate either deletion or preservation.

Common mistakes about keeping recordings after transcription

Use transcription as a point to review retention, not as an automatic deletion date

Once transcription and initial verification are complete, ask why the original identifiable recording still needs to exist. A clear answer may justify retention. The absence of one should prompt a closer look at whether continued storage is necessary.

Document the decision and the reason behind it. "Delete after transcription," "retain for ten years," and "retain indefinitely" are not self-justifying policies. The retention period should follow from the purpose and applicable requirements.

Before deleting or retaining the original recording

Questions about retaining research recordings

Transcription should trigger a retention decision, not predetermine it

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.