AU files appear in early workstation, Unix, education, and preservation collections. The format can carry several encodings, so the ingest worker checks the real stream rather than equating audio/basic with one guaranteed waveform.
Where this workflow fits
- Unix audio collections
- Academic sound archives
- Legacy workstation recordings
Read the legacy audio header
The Sun/NeXT AU structure begins with a compact header describing offset, size, encoding, sample rate, and channels. A valid suffix alone cannot prove those fields or the following sample data are internally consistent.
Keep provenance beside the transcript
Preservation audio benefits from a simple audit trail. Note the source identifier, inspect uncertain words against the untouched file, and keep any normalized derivative separate from the collection master.
Before you upload
- Preserve the archival original and document any conversion performed before transcription.
- Compare duration and audible content with the collection record to detect truncation.
Limits worth knowing
- Historic AU material may use low sample rates or companded audio with limited speech detail.
- Damaged or unusually annotated headers can prevent a decoder from locating the audio payload.
Questions from the workflow
Should an AU header or annotation be rewritten before upload?
Keep the untouched AU master. If an annotation or unusual data offset prevents decoding, create a documented derivative, record the tool and settings, and compare its duration and checksum-tracked provenance with the original.
Reviewed sources
- IANA Media Types registry · checked 2026-07-20
- RFC 2046: basic audio media type · checked 2026-07-20