Recording Oral History, So the Voice Survives

This is Talk Type, from the team at Pepys, where we turn talk into text.

Think about what you're actually trying to save. Not just facts. A voice. The way your grandmother pauses before she says a name. The story only she can tell, in the words only she uses. Oral history is that, captured before it's gone.

And the recording is the thing. Not the transcript. The audio is the archival object, and the transcript sits beside it. So capture it to standard. The sound and audiovisual archives standard, IASA-TC 04, calls for uncompressed WAV or BWF, at least forty eight kilohertz, and advises twenty four bit depth. MP3 doesn't meet that. Record lossless from the start, because you can't recover the detail a lossy file threw away.

No tool transcribes what the microphone missed. Get the recorder close to the narrator, off any hard surface that booms, away from the vents and the ticking clock. Then open with a slate. Say it out loud. Your name, the narrator's name, the date, the place. That one habit anchors which voice is which, and timestamps the session before the first question.

There's paperwork, and it matters more than it looks. The Oral History Association points to a deed of gift, the document that transfers rights. Copyright here is shared by default. An oral history is co-created, so it usually belongs to both the interviewer and the narrator until a signed agreement moves it. The narrator owns the words they speak. Transferring ownership and opening the interview to researchers are two separate decisions.

Now the labels. Oral history has its own convention. Not Speaker One and Speaker Two. Narrator, for the person telling their story. Interviewer, for the person asking. Get a diarized draft that splits the two voices, then rename them once, up front. A later researcher should be able to tell who's talking without playing the tape.

Here's the part people underestimate. The transcript isn't typed from a blank page. It's an audit. You listen against the recording, line by line, catching the wrong words, the homophones, the misheard names. Researchers found manual transcription can run up to about six hours per hour of audio. Start from a labeled first pass, and most of that turns into a focused check instead of a slog.

Be honest about the hard spots. Where the audio's uncertain, type your best guess and flag it. Where you can't guess, leave the blank and flag it anyway. Keep those marks in the final copy. A flagged uncertain word is honest. A confident wrong word becomes a misquote in someone's research.

Then the voice gets the last read. The narrator reviews the transcript. They can correct it, annotate it, restrict it, even withdraw it, before it's shared. Their notes get bracketed in, so the addition reads as theirs and not the editor's.

The sessions add up. A life story rarely fits one sitting, and the hours stack across visits. So you want a tool that takes any length, no thirty minute wall. You pay once for what you use. You keep the real file. And your family's audio is never used to train anything.

Deposit the corrected transcript with the master audio and the signed deed. Because a hundred years from now, someone will want to hear it in her own words.

That's this episode of Talk Type. The full write up, with the links and sources, is in the show notes. Pepys transcribes any file or link, any length, pay once, and we never train on your audio. Your first sixty minutes are free at pepys dot co. Thanks for listening, and we'll see you next time.