Recording Oral Histories: Capture, Convert, and Archive

An oral history project — recording elders in a family, veterans in a community, longtime workers at an institution — carries a peculiar responsibility: the recordings must be listenable this year and playable in fifty. Those are different engineering problems. Listenability wants small, convenient files that play on a phone in a kitchen; longevity wants formats chosen for the archive, stored redundantly, documented well enough that a stranger can make sense of them decades on. The solution is not to compromise between the two but to produce both: a lossless archival master and an MP3 access copy of every session.

Capture: get it right once

Everything downstream depends on the recording session, so a few capture rules are worth stating even though this is mostly a conversion guide:

• Record WAV if the device allows it — a dedicated recorder or a phone app that supports lossless capture. 44.1 or 48 kHz, 16- or 24-bit; the sample rate FAQ explains why either is fine for voice.
• Quiet room, close microphone. Refrigerators, clocks, and traffic are forever once recorded.
• Let people finish. Long pauses are part of testimony; do not stop the recorder during them.
• Say the metadata out loud. Open each session by stating the date, place, narrator, and interviewer on tape — the one label that can never be separated from the recording.

The two-output workflow

The archival master

The master is the original-resolution, unedited capture, and the rule is simple: never touch it. Store the WAV exactly as recorded. If your recorder produced only a compressed format, keep that original file as the master anyway — it is the closest thing to the event that exists. For long-term storage, FLAC is the archivist's friend: it is lossless, open, and typically 40 to 60 percent of WAV size, which matters across a hundred hours of interviews. The FLAC FAQ covers why archives standardized on it, and the broader lossy vs lossless explainer is worth reading before you choose. If a session ever exists only as MP3 — a narrator's old recording, a rescued file — an MP3 to FLAC conversion will not restore lost quality, but it does freeze the file against any further generational loss in future copies.

The access copy

The access copy is the file people actually use, and for that MP3 remains unbeatable: every device, every platform, every relative's phone. Convert each master with a WAV to MP3 converter at 128 to 192 kbps — speech is transparent well below that, and a two-hour session lands near 100 MB, small enough to email in parts, stream from shared storage, or copy to a USB stick for a narrator who wants their own recording back.

Processing the access copies

The master stays raw; the access copy earns some gentle processing:

• Normalize. Elderly voices are often quiet, and level varies between sessions recorded months apart. A loudness normalization pass makes the whole collection playable at one volume setting — see the normalization FAQ for what this does and does not change.
• Trim setup noise. Cut the microphone bumps and chair scrapes before the spoken introduction with a trim tool — but keep the spoken date-and-name slate itself.
• Cut excerpt reels. A 90-minute interview is for the archive; a six-minute reel of its best stories is what family members actually play at gatherings. Merging the strongest excerpts across sessions into a single listening reel is often the most-appreciated deliverable of the whole project.

Metadata: the part everyone regrets skipping

A recording nobody can identify is barely an archive at all. Three layers, in order of durability: the spoken slate at the top of the tape; the filename ("2026-04-02 eleanor-vance session-3.mp3"); and a plain-text index file per collection listing narrator, dates, topics, and timecodes of notable stories. Embedded ID3 tags in the MP3s help media players display titles properly — the ID3 FAQ explains what they store — but treat tags as a convenience, not the record: the text index is what survives format migrations.

Returning recordings to narrators

Plan from the start to give every narrator their own recording — it is both good ethics and the gesture that keeps families engaged with the project. The MP3 access copy is the right artifact for this: small enough to email or copy to a USB stick, playable on whatever device the narrator or their family owns, and already normalized so it plays comfortably. If a narrator's family wants everything, hand over the full session MP3s plus the excerpt reel; the reel is invariably the one that gets shared onward at reunions and, eventually, memorial services. This is also the quiet argument for doing the access-copy processing well: the file you hand back is, for many families, the version that becomes the recording.

Storage: the 3-2-1 floor

Three copies, two different media, one off-site. In practice for a small project: the working copy on a computer, a mirror on an external drive, and a cloud copy. Add a fourth consideration unique to oral history — succession. Someone other than you should know where the archive lives and how it is organized, because fifty-year archives outlive their creators' hard drives, and sometimes their creators.

A session-day checklist

• Record WAV; slate the session verbally.
• Copy the card to two locations before deleting anything.
• File the untouched master (WAV or FLAC).
• Convert to MP3, normalize, trim; file the access copy beside the master.
• Update the index file the same day, while names and stories are fresh.

The narrators are giving you the scarcest thing they have — their memory, in their voice. The dual-format workflow costs perhaps fifteen minutes per session, and it is the difference between a shoebox of mystery files and a collection a great-grandchild can open, understand, and hear.