How to Transcribe a Family History Interview

How do you transcribe a family history interview?

Transcribe a family history interview by listening in short sections, identifying each speaker, and checking the entire draft against the recording. Mark unclear words instead of guessing, and separate later corrections from what was actually said. Agree on storage, outside transcription services, review and sharing permissions before proceeding. A transcript makes a conversation easier to search; it does not replace the person's recorded voice.

This guide begins after the interview. For preparing the conversation itself, use our family history interview questions. The workflow and examples below are Keepsake Map's suggested home-project system, not a universal archival template.

Do you need a transcript, a summary or a timed index?

These are different deliverables. A transcript represents the spoken conversation in text. A summary describes its main subjects in the writer's own words. A timed index points to topics at particular places in the recording.

The National Park Service's transcribing guidance treats transcripts, indexes and abstracts as options whose usefulness depends on the project. Decide what you need before committing to word-by-word work.

For a relative who wants to find a story about the family shop, an index might be enough. For someone who wants to read the conversation, a summary is not an equivalent substitute.

An original, entirely fictional index could look like this:

The school section starts 2 minutes 15 seconds into this invented recording, not at 2:15 in the afternoon. The interval from 02:15 to 07:40 is 5 minutes 25 seconds. Label positions consistently so a future listener can actually find the passage.

What should you set up before typing?

Check the existing permission record. Do not assume agreement to a family recording includes uploading it to a transcription company, distributing a searchable transcript or posting it publicly.

The Oral History Association's ethics statement calls for informed consent and an opportunity for narrators to review and approve material before public release or archival deposit. Discuss intended uses and limits, including options to restrict, clarify or remove material. Do not promise control over every later reuse once something is public. Obtain qualified advice if rights or legal restrictions are unclear.

For the working folder, our suggested cover sheet has these fields:

• Interview ID and recording filename.
• Interview date and participant labels.
• Transcriber's name and draft date.
• Recording duration, copied from the player.
• Draft status: unchecked, audio-checked, narrator-reviewed or approved for a specified use.
• Location of the permission record and name of the person managing access.

An ID such as OH-014 connects the pieces without putting a person's full name in every filename. It is an organizational label, not an anonymity safeguard.

Use separate filenames such as OH-014-transcript-v01 and OH-014-transcript-v02. Keep the audio as a separate item under the agreed retention and access arrangements. Saving a text document with the same interview ID does not make it an audio backup.

How much should you tidy the speech?

Decide before you begin. Otherwise the first page may preserve every hesitation while the last turns into polished prose.

Our suggested house rule is to keep the speaker's wording and meaningful repetition, add readable punctuation, and identify editorial additions. Do not rewrite grammar to make someone sound more formal. Explain any routine omissions, such as repetitive interviewer acknowledgments, in a short editorial note.

Baylor University's oral-history resources emphasize that a transcript cannot capture everything in an interview and that listeners should return to the recording. Tone, pacing and interaction are reasons to keep text and audio connected.

Choose a small notation key. For this home-project example:

• Speaker labels begin each change of speaker.
• [unclear, 00:06:18] marks a passage the transcriber cannot resolve.
• [laughs] records an audible action when relevant to the meaning.
• [Editorial note: ...] introduces later explanation, not spoken words.

These are our proposed conventions, not a claim that every institution uses identical symbols. If an archive will receive the interview, ask for its requirements before formatting the whole document.

Avoid interpretations such as [angrily] when all you can hear is a louder voice. A transcript should not quietly add a motive.

What does a careful transcript passage look like?

Here is a wholly invented teaching example, not a quotation from a real interview:

Interviewer [00:04:10]: When did you leave the shop?

Narrator [00:04:14]: In 1968. No, wait, that was when we moved. I think I left the following spring.

Interviewer [00:04:24]: Was the shop on Mill Road?

Narrator [00:04:27]: On [unclear, 00:04:28] Road, beside the old cinema.

Do not turn that into "The narrator left the shop in spring 1969 on Mill Road." That combines an inference about the year with an unverified place name and discards the speaker's uncertainty.

In the working query list, record two separate questions: "Can the road name at 00:04:28 be clarified?" and "Is there independent evidence for the leaving date?" Answering the first is a transcription task. Answering the second is historical research.

Baylor's 2018 transcript style guide distinguishes added editorial material from recorded speech and discusses retaining meaningful false starts and repetitions. That distinction protects the reader from mistaking a later explanation for the original account.

If the narrator later supplies a spelling, date that clarification and identify its author. If a document supplies a different date, cite the document in a separate research note. Neither should silently overwrite the recording's words.

Can automatic transcription do the first pass?

It can produce a draft, but a fluent paragraph is not proof that it matches the audio. The OHA archiving manual describes automated transcription as a possible rough first step and says both human and automated first drafts need accuracy review, including names and dates.

Before any upload, obtain permission for that specific outside processing. Our editorial precaution is to read the service's current terms for storage, access, retention and reuse; if those arrangements cannot be understood or accepted, do not upload the interview. This guide does not endorse or audit a service.

Check the whole draft while listening, rather than sampling only its opening. Keep a separate query list for unresolved passages. Search for each confirmed name afterward to catch inconsistent spellings, then check the corrections in context.

For our fictional OH-014 project, use a simple review record:

A checked spelling is not proof that an event happened. Keep those questions separate even when one person is doing both jobs.

What needs checking before sharing?

Review the proposed share copy, not just the earlier working draft. Confirm that its permission status applies to the intended audience and format. A private family copy and a public website are different uses.

Changing a name does not guarantee anonymity. The OHA's definition of anonymous explains that a pseudonym can leave someone identifiable through other information. Occupation, relationships and a distinctive event may still reveal who is speaking. Consider the whole account and people mentioned in it.

Our final-release checklist asks whether restricted passages appear in the transcript, attached audio, filename, summary or accompanying photograph caption. Do not assume removing words from one document removes them from the rest of the package. For sensitive redaction, ask a suitably qualified archivist or information-security professional to check the release process; a visual covering alone is not an assurance of removal.

If permission changes, pause distribution and agree how the retained originals, working copies and backups will be handled. Do not promise that copies already passed to other people can all be recovered.

How do you keep the finished work usable?

The Library of Congress's personal digital-audio guidance recommends descriptive organization, multiple copies in different locations and periodic checks that files remain readable. Apply that within the interview's agreed privacy and retention limits.

Our suggested handover note lists the approved transcript version, corresponding recording, current restrictions, backup locations and responsible contact. Open the actual backup files when checking them; a folder name alone tells the next custodian little.

A practical completion test is straightforward: can an authorized reader open the approved transcript, locate a passage in its matching recording, distinguish spoken words from later notes, and understand what may be shared?

For photographs discussed in the interview, connect their existing identifiers using our family photo organization guide. Keep the transcript's task modest and useful: make the conversation findable without substituting a cleaner story for the one actually told.

Sources

• National Park Service: Interview Transcribing.
• Oral History Association: Statement on Ethics.
• Baylor University: Institute for Oral History Resources.
• Baylor University: Transcript Style Guide, revised May 2018.
• Oral History Association: Archiving Oral History Manual.
• Oral History Association: Anonymous.
• Library of Congress: Preserving Your Digital Audio.