Export & import¶

Formats¶

Format	Content
`md`	YAML frontmatter (`title`, `date`, `duration`, `type`, `rex_id`, `model`, `speakers`, `organizer`, `attendees`, `tags`), an `[Open in Rex](rex://open?id=…)` link, then `[HH:MM:SS] Speaker: text` - the same renderer as `transcript.md`. `organizer`/`attendees` appear only when Teams attendance enrichment resolved (see `[attendance]` in Config).
`txt`	Plain text, one segment per line
`json`	`transcript.json` verbatim (the public schema, with words and confidence)
`srt`	Numbered cues, `HH:MM:SS,mmm`, `Speaker: text`
`vtt`	WebVTT with `<v Speaker>` voice tags

A segment whose speaker is unknown is written without a label.

Exporting¶

Toolbar › Export… lists every meeting with a transcript (the selected one preselected), the formats from [export] formats, an include audio toggle and an optional Export into one folder destination.
Without a destination each meeting gets <meeting>/exports/<slug>.<ext>; with one, everything lands there as <yyyy-MM-dd>-<slug>.<ext>, never overwriting (-2, -3, …).
[export] auto = true exports after every successful transcription, before the transcript-ready workflows run.
CLI: rex export <folder…> [--format md,srt] [--audio] [--to <dir>].

Importing¶

Drag any audio file (wav, m4a, mp3, aac, caf, aiff, flac) onto the main window. Rex probes the file without decoding it and asks for a title and, when the file has two sides, which side is you:

File	Choice
mono	none - everything becomes them (no AEC)
stereo	First side is you / second side is you / everything is others / everything is you
two tracks (a Rex recording)	same, defaulting to mic,system

The file is copied into a new meeting folder (source.kind = import), the chosen mapping is stored in meeting.json.recording.tracks, and transcription is queued. The original is never touched.

CLI: rex import <audio> [--title …] [--layout auto|mic-system|system-mic|system|mic] [--no-transcribe]. auto means two tracks → mic,system; anything else → system.