10 March 2026 | 6 min read
Metadata workflow: from card to searchable library in one pass
Doing it once properly costs twenty minutes per card. Doing it later costs a weekend and you will still get half of it wrong.
Ingest is the only moment when you remember what you recorded. Everything about a good metadata workflow is designed to exploit that window before it closes.
Step one, back up before touching anything
Two copies on two devices before a single file is renamed, moved or converted. Everything below is reversible only because this step happened.
Step two, inspect a sample
Check one file per card: sample rate, bit depth, channel count, track names, timecode present or not. If the recorder was misconfigured, you want to know now, not after renaming three hundred files around a wrong assumption.
Step three, split and convert
Split poly WAVs into named mono stems if your library is mono based, and convert anything exotic into the format you actually store. Do it in a tool that preserves the embedded names, otherwise you are handing yourself a renaming job.
- Keep the sample rate of the source unless you have a delivery reason to change it.
- Stay at 24 bit for library material; 32 bit float is for capture, not storage.
- Never route unreleased material through an upload based converter.
Step four, name and tag together
Apply the naming convention and write the same descriptive words into the description field in the same sitting. Doing them separately guarantees they drift apart.
Step five, verify and file
Spot check the output, then move the files into the library and mark the card as ingested. An explicit ingested marker is what stops the same card being half processed twice.
Everything in steps two and three can be done here in the browser: inspect, split, merge, map channels and export at the rate, bit depth and format you store in, with no file leaving your machine.