What oral history recordings look like
Oral history projects collect long, personal interviews. A local history society might record elders on a handheld recorder that saves WAV. A family project may be a grandparent talking into a phone, saved as M4A. Older collections arrive as digitised cassette or reel-to-reel tapes, usually as WAV files from the transfer.
All of these can be uploaded as they are. Large WAV files take longer to send, and the upload is resumable: if it stops, Retry continues from where it left off. Video interviews in MP4 or MOV work too, with the audio extracted on our side.
Recording and digitising tips
Narrators are often older and may speak softly, so microphone distance matters more than usual. A clip-on lavalier mic on the narrator, or a recorder close to them on a soft cloth, gives much better results than a recorder across the room. Switch off clocks, fridges and televisions where you can.
For tapes, the transfer sets the ceiling. Clean the tape heads, check the playback speed and avoid heavy noise reduction during transfer. Aggressive filtering can make voices sound watery, which hurts recognition more than gentle hiss does. Keep the original transfer as your archive master and upload a copy.
Correcting names, places and dialect
Oral histories are full of the things speech recognition knows least: family names, small villages, old company names, local slang and dialect words. Expect to correct these. The synced player helps: the current line is highlighted as it plays and you can click any line to hear it again. Edit the text in place as you go.
Many narrators switch languages or speak with strong regional accents. More than 90 languages are supported, and you can set the language per file instead of relying on detection. Mixed Chinese and English speech is handled. Heavy accents, very quiet speech and worn tapes lower accuracy, so plan for a careful listening pass on archive material.
If you have speaker identification (paid plans or a Minute Pack, not the Free plan or trial), rename the speakers to "Interviewer" and the narrator's name. Family recordings with several relatives chiming in benefit most from this.
Exports for archives, families and researchers
Choose exports by who will use them. You can turn timestamps and speaker names on or off for each one.
Many oral history programmes publish a time-coded index next to each interview so visitors can jump to a topic. Keep timestamps on in your working copy and you can build that index from the transcript, noting the time where each subject starts. The CSV export puts every segment in its own row with its start time, which makes sorting and copying into a catalogue spreadsheet easier.
- DOCX: for editing into a final transcript with your project's header, release details and abstract.
- PDF: a fixed copy to deposit with an archive or print for the narrator's family.
- TXT: plain text for catalogue systems and long-term preservation alongside the audio file.
- VTT or SRT: captions if you publish interview clips as video online.
- Read-only share link: let the narrator or relatives read the transcript and send corrections.
Practical limits
Oral history sessions often run for several hours. Free accounts have a per-file length cap, and paid plans allow much longer files and batch uploads for a whole collection. Use folders to group interviews by narrator or project.
Many archives publish transcripts with a review by the narrator. An automatic transcript is a strong first draft for that process, but it does not replace it. Files are never used to train AI models, which matters when the stories are private, and you can delete recordings whenever you choose.