Library size, transcription status counts, worker health and configuration.

List memos, newest first.

Args: limit: page size (max 200). offset: start index for paging. since: only memos recorded on or after this date (YYYY-MM-DD). until: only memos recorded on or before this date (YYYY-MM-DD). title: case-insensitive substring of the title. status: transcribed, pending, segmenting (longer than the engine accepts: being transcribed in parts), not_transcribed, not_downloaded, failed or too_long (long and cannot be split: splitting off or ffmpeg missing).

Read one transcript with uncertainty marks.

Args: memo: memo id from voicememos_list (8 characters is enough). format: 'markdown' (with header), 'text' (transcript only) or 'json' (word-level data). threshold: confidence threshold for marking words (default from config); re-renders without re-transcribing. offset: character offset for long transcripts (see next_offset). max_chars: maximum characters returned (1000-100000).

Search transcripts and titles (case-insensitive). Returns snippets per memo.

Args: query: text to find, or a regular expression when regex=true. limit: maximum number of memos returned. since: only memos recorded on or after this date (YYYY-MM-DD). until: only memos recorded on or before this date (YYYY-MM-DD). regex: treat query as a regular expression.

Sentences the user tagged by starting them with a spoken marker word (e.g. "Idea: ...").

Args: since: only memos recorded on or after this date (YYYY-MM-DD). until: only memos recorded on or before this date (YYYY-MM-DD). label: only this marker label. limit: maximum number of markers returned.

Queue memos for transcription in the background. Returns a job_id immediately.

Args: memos: memo ids to transcribe. since: or: all memos recorded on or after this date (YYYY-MM-DD). until: or: all memos recorded on or before this date (YYYY-MM-DD). redo: transcribe again even if a transcript exists. push_roam: also append the transcripts to Roam daily notes (if Roam is configured).

Queue a second, independent reading (Gemini) of already transcribed memos.

Disagreements with the primary engine appear as {a|b} in the transcript. Returns a job_id.

Args: memos: memo ids. since: or: memos recorded on or after this date (YYYY-MM-DD). until: or: memos recorded on or before this date (YYYY-MM-DD). model: Gemini model id (default from config, e.g. gemini-4-argon). redo: read again even if a second reading exists. limit: at most this many memos (oldest first). only_uncertain: only memos with low certainty; long recordings (read in parts) only in their parts with low certainty. realign: no new reading; align the stored readings again with the current rules.

Compare primary engine and second reader per language (agreement, calibration, samples).

Writes a Markdown report with a listening checklist and returns the summary.

Args: since: only memos recorded on or after this date (YYYY-MM-DD). until: only memos recorded on or before this date (YYYY-MM-DD). samples: disagreements to sample per language (max 50).

Recognise the owner's voice with a local voiceprint.

Args: action: 'status' (enrollment state, thresholds and suggested enrollment memos), 'enroll' (build the voiceprint from memos where only the owner speaks), 'identify' (apply it to already transcribed memos), 'add_other' (store a known other voice, e.g. a relative who sounds similar, from one speaker of one memo; it is never shown by name, it only prevents mistaking that person for the owner) or 'remove_other' (by name). memos: memo ids for 'enroll' (2+, ideally 4-6 from different years), 'identify', or one memo for 'add_other'. since: for 'identify': memos recorded on or after this date (YYYY-MM-DD). until: for 'identify': memos recorded on or before this date (YYYY-MM-DD). name: the owner's name for 'enroll' (default from config); the other person's name for 'add_other' / 'remove_other'. speaker: for 'add_other' in a multi-speaker memo: the speaker label (S1, S2, ...).

Progress of a transcription job, or the 10 most recent jobs when job_id is omitted.