v0.19.1

AutoDJ 0.19.1 fixes one indexing problem. Full details are in the CHANGELOG.

Fixed

• Beets entries for deleted files no longer fail every index run. When the beets library still listed files that were gone from disk, each autodj index pruned them and then tried to embed them again, and every one failed. Indexing now skips beets items whose file is missing and logs one warning with the count, a few example paths, and how to clean the beets library (beet update -M, or beet update -p to preview).

No configuration or data changes. Upgrading from 0.19.0 needs only the new code.

v0.19.0

AutoDJ 0.19.0 is a large simplification release: about 45,000 lines removed, no backward compatibility, and many fixes found by an NVDA release sample and a fresh-clone walkthrough. Full details are in the CHANGELOG.

Before you upgrade

Nothing old is converted automatically. Read the CHANGELOG "Removed" and "Changed" sections, then run uv run autodj doctor after upgrading.

• No web build step. The server serves src/autodj/static directly. You no longer need Node or npm run build to run AutoDJ; delete any leftover src/autodj/static_dist folder.
• Index format. Indexes from 0.18 use manifest schema 2 and are refused with a rebuild message. Rebuild with uv run autodj index --force. The tracks database and vectors themselves did not change, so the rebuild re-embeds your library.
• Model folder. The model is now stored in the Hugging Face cache layout under model_dir, so the first autodj index downloads it once more. You can delete the old model folder afterwards.
• Removed: autodj play (use autodj serve --server-audio), the DJ-mix options on serve (use config.toml or Settings), --export-m3u, --history-file and autodj playlist, Daypart and Mood arc, the Entropy walk pick mode, the preset genres filter, presets.toml (presets live in config.toml), backup --online, and the non-worklet effect fallbacks.
• Stricter config. Unknown keys and sections, and out-of-range values, now stop AutoDJ with an error that names them. Profiles and backups from earlier versions are refused.

Highlights

• HTTPS from config ([server] ssl_certfile and ssl_keyfile), with renewed certificates picked up without a restart, and a guide to using your own domain with a Cloudflare Tunnel and Cloudflare Access.
• Paired devices in Settings: rename and revoke devices, and show the pairing code to pair a new one.
• Picks stay musical: silent tracks are never picked automatically, and sets follow Similarity (optionally anchored to the seed).
• Serato cues are imported from audio file tags, and imported cues now reach indexed and player-analysed tracks.
• Simpler backup and restore, simpler liner uploads that work in group-writable folders, and one beat-grid method on every machine.
• Screen reader and safety fixes: shortcuts do nothing while a dialog is open, Play now works on a fresh page, nothing ever plays without your own Play, every sound follows the page volume, and many announcements and labels were corrected.

Accessibility testing

The release sample is in docs/accessibility-samples/0.19.0.md: NVDA 2026.2 with Chrome (all flows), Firefox and Edge (a subset), on Windows 11 over HTTPS. An automated agent pressed the keys and captured NVDA's speech; no person listened. The defects it found were fixed before this release. Known quirks, both from how NVDA and Chrome handle dialogs: a dialog's name is read twice when it opens, and one stray line is read when a dialog closes.

v0.18.2

AutoDJ 0.18.2 fixes two playback problems. Full details are in the CHANGELOG.

Fixed

• Silent tracks no longer get picked automatically. Libraries often contain short "[silence]" gap tracks. They were eligible for every automatic pick, and the "Entropy walk" pick mode (which deliberately picks the least similar track each time) landed on one every other track. Automatic picks now skip tracks whose measured level is digital silence (below about -80 dBFS); you can still play one yourself from search. If your sets wander away from the seed, set Settings, Pick mode to "Similarity (default)", and tick "Anchor to seed" to stay close to it.
• A saved Discovery rate now works after a restart. The rate was restored but discovery stayed off until you touched the setting.

No configuration or data changes.

v0.18.1

AutoDJ 0.18.1 fixes two problems found while listening to 0.18.0. Full details are in the CHANGELOG.

Fixed

• Play now from search did nothing on a freshly opened page. The server switched to the chosen track, but the page stayed silent until you pressed Play on Now Playing. Play now on a search result now starts playback in that page, at your volume, and says "Playing ... now." once, only after sound has started. Play next, Add to queue, server updates and media keys still never start a page on their own.
• Keyboard shortcuts still worked behind open dialogs. For example, pressing n while "Clear queue?" was asking skipped the track. Shortcuts now do nothing while any dialog is open.

No configuration or data changes. Upgrading from 0.18.0 needs only the new code and web bundle.

v0.18.0

AutoDJ 0.18.0 adds a radio stream for Sonos, VLC and other players, removes every backward-compatibility path, and brings a large round of screen-reader fixes. The full list is in the CHANGELOG.

Before you upgrade

This release removes old formats instead of converting them. Read the "Removed" and "Changed" sections of the CHANGELOG, then run uv run autodj doctor after upgrading.

• config.toml and config.local.toml now reject unknown sections and keys. Delete any harmonic_mixing line and use harmonic_mode (off, compatible, strict, energy_boost, mood_change or neighbour). Leftover path_remap or playback.dayparts_dir lines also stop startup.
• web_state.json files from earlier versions are ignored, so saved web settings go back to their defaults until you save them again.
• Indexes made before the manifest format, or indexes that store absolute paths, are refused. Rebuild them with uv run autodj index --force.
• serve --no-playback, index --reindex-modified-since, the -a/-e/--analyse/--enrich index flags, the autodj[web] extra, and the "Weighted" liner mode are gone.
• Server-side mixing (--server-audio and --stream) skips tracks longer than [playback] server_max_track_minutes (default 15) so a small machine does not run out of memory.

Highlights

• Radio stream (serve --stream): an MP3 stream with track titles for Sonos, VLC and other network players, including the now-playing title.
• Voice liners and the server mix bus also work with plain --server-audio.
• Web page volume now covers everything: transition effects, voice liners and the stream player follow the volume and mute. Effects stay at or below the music's level. The page never starts playing on its own, including after a server restart or from a media key, and the server remembers its volume.
• Screen-reader work: announcements from every tab (opt-in), spoken results for hotkeys, new status keys (Shift+E, Shift+V, Shift+Q, Shift+J, Shift+L), keys spoken in words ("F sharp minor"), plain-language errors, a line-by-line job log, and symbols given words at NVDA's lowest punctuation level.
• New in the web page: Profiles, paired devices and sign-out, Play next, Clear queue, Move to top, running Analyse, more mix settings, and a History cleanup.

Accessibility testing

The release sample is recorded in docs/accessibility-samples/0.18.0.md. It was taken with NVDA 2026.2 on Windows 11, with Chrome for all flows and Firefox and Edge for a subset. An automated agent pressed the keys and captured NVDA's speech; no person listened. Narrator was not sampled.

Known issues, both caused by how NVDA and Chrome handle dialogs:

• NVDA reads the pairing dialog's name and instructions twice when it opens.
• When a confirmation dialog closes, NVDA reads one line from the top of the page before it reaches the next control.

v0.16.1

Full Changelog: v0.16.0...v0.16.1