Skip to content

Library, subscriptions, and queues

The authoritative library stores episode records, Markdown, groups, and state. Subscriptions supply entries; processing creates reading documents. Audio/work caches remain on the transcription computer.

Subscription workspace with counts, unprocessed filters, selection, and Add Subscription.

Real current macOS source build. Counts describe the example library, not capacity/accuracy claims. Full image.

Add and refresh

Open Subscriptions, paste RSS/Xiaoyuzhou podcast/Apple Podcasts show into Add Subscription, and choose a group. Refresh updates metadata, not an automatic audio download or complete-series transcription.

atot exec subscriptions add "https://example.org/podcast/rss.xml" \
  --group-name Research --limit 10 --json
atot exec subscriptions list --json
atot exec subscriptions refresh --all --json
atot exec subscriptions refresh --id <subscription-id> --full-history --json

Ordinary CLI import/refresh defaults to 10 entries. Use the returned ID. Full history includes only entries currently exposed by the feed. Refresh is manual; external schedulers can invoke CLI, without a background-polling promise.

Filter and batch

Use totals, Unprocessed only, and time filters, then Select all/Deselect all, current-source First N, or individual rows.

  1. Pick feed and filters.
  2. Select/check a limited batch.
  3. Choose local-only work without cloud keys.
  4. Wait for results and review QA before the next batch.

Selection does not start processing and some items may be ineligible. Audio pending needs media; old recordings may no longer be available.

Organize and read

Groups organize topics/projects, Starred marks important episodes, and read/archive state manages what to revisit. Today/Unprocessed/group views are not duplicate documents. Group changes do not invoke ASR; read state does not prove verification.

Overview, Transcript, and Summary stay separate. Missing notes need MCP or cloud, not just opening a tab.

Delete or unsubscribe

Unsubscribe stops managing the feed. Episode deletion removes its record/documents and writes a tombstone against silent reimport. Back up needed documents first.

atot exec subscriptions remove <subscription-id> --json
atot exec library episode delete <episode-id> --json

Do not remove internal JSON to force refresh. Tombstones belong to sync/deletion state; back up the complete library.

Share the same library

Settings → Library selects local/iCloud. Default iCloud:

~/Library/Mobile Documents/com~apple~CloudDocs/Audio2Text/Audio2TextLibrary/

Local:

~/Documents/SpeechToTextData/Audio2TextLibrary/

These are different authoritative locations. Weights/media/work remain local; switching locations does not prove that both have identical data. Verify/back up before moving.

iOS Files selects the existing Audio2Text or Audio2TextLibrary folder. It reads/manages documents and subscription metadata, without ASR/audio processing. See iOS.

Check actual data

atot exec library status --json

Compare resolved paths/storage mode, filters, and iCloud file availability when clients disagree.

Workspace → · Backup → · Sync troubleshooting →