Markers and Labels

Add a named point marker or region annotation to the timeline.

Example prompt: mark the chorus at 1:05

Returns: annotation_id

Import Audacity-format label text into the session as annotations. Format: each line is 'start_sec TAB end_sec TAB name'.

Example prompt: import these Audacity labels

Returns: node_id

Export session annotations as Audacity-format label text (start_sec TAB end_sec TAB name, one per line).

Example prompt: export the markers as labels

Returns: label text

Find markers and regions by name (any case, part of it) and/or time range, in time order, with the ids edit_markers and delete_markers take. Regions that overlap the range count.

Example prompt: list the markers named chorus

Returns: text, summary, next_offset

Read-only, so Plan first never holds it. Pages like list_session: at most 4 KB or 100 lines a call, with next_offset for the rest.

Add several markers or regions in one step: a list of {name, time} or {name, start, end}, or a marker every N seconds from a start up to and including an end (default the session's end), named from a template such as "Chapter {n}".

Example prompt: put a chapter marker every 5 minutes

Returns: node_id, added, summary

At most 500 a call; an interval that would place more is refused before any is. The end gets a marker when it falls on a step (every 600 s to 3600 s is seven). A list takes no from, to or name: those belong to the interval form, and the call is refused rather than ignoring them. One undoable step.

Rename, move or resize markers by id. time moves one (a region keeps its length); start and/or end set a region's edges.

Example prompt: rename the second marker and move it to 1:30

Returns: node_id, edited, summary

All or nothing: an unknown id or an impossible region changes nothing. One undoable step. The label lane uses the same rules.

Delete markers by id, or every marker matching a filter (name, from, to). A filter needs at least one field; ids and a filter together are refused.

Example prompt: delete the markers between 2:00 and 3:00

Returns: node_id, deleted, summary

A name is part of a name, any case ("intro" also takes "Introduction"). Unlike find_markers, a from/to range takes only labels wholly inside it, so a region that merely overlaps or spans the range stays. One undoable step; undo brings them back.

Add point markers at the starts of the stored transcript's segments (default) or sentences (mode). A file or unknown-source transcript needs an explicit offset_sec, even 0. Each label is the first six words, at most 80 characters.

Example prompt: add a marker at each transcript sentence

Returns: node_id, marker_count, skipped_count, time_basis, time_note, source, applied_offset_sec, summary

Run transcribe first. Times follow edits that move, cut or stretch the track; align_to_beat leaves them stale. A marker with the same name and time (within 1 ms) is skipped, and a run that adds none appends no node. One undoable step; no audio changes.