Settings reference

Settings reference

Settings is a brass-edged section hub of cards in Settings → (section) — currently 13 cards (post-v0.5.42 redesign, up from 8). Settings is a primary nav destination since the v0.5.40 nav restructure ({Playing · Library · Voices · Settings} four-tab dock, finalized v0.5.50).

Settings hub cards

Voice & Playback

• Default speed — 0.5×–4× slider (#194). Per-voice defaults persist (#195).
• Default pitch — ±20% slider. Per-voice defaults persist (#195).
• Azure — BYOK setup for Azure Cognitive Services HD voices (#182–#186).
  • Key (paste from Azure portal)
  • Region (e.g. eastus, westeurope)
  • Test connection — round-trips the synthesis API to validate
  • Offline fallback voice — local voice to use if Azure errors mid-chapter

Voice library

Top-level entry post-v0.5.42 (no longer nested inside Voice & Playback). Engine-grouped picker — Piper (45), Kokoro (53), KittenTTS (8), Supertonic 3 (10), and — once configured — Azure HD (20). ★ favorites surface to the top. Supertonic 3 (~139 MB shared model) is the newest local family — live since v1.2.3 (see Voices).

Reading

• Theme — System / Light / Dark.
• Sleep timer — durations + end-of-chapter mode.
• Shake to extend sleep timer (#150) — when ON, shaking during the 10s fade tail re-arms the timer for 15 minutes. Disable for bumpy commutes.
• Auto Do Not Disturb (#1190 / #1199, v1.2.0) — optionally silence notifications while the sleep timer is running.
• Bedtime auto-sleep (#1110, v1.1.4) — arm the sleep timer automatically when your phone enters Sleep / Bedtime mode (Samsung Modes & Routines, Android Digital Wellbeing, or manual DND), at your last-used shake-extend duration. Toggle under Voice & Playback.

Performance

• Warm-up Wait — ON shows a brass spinner during engine cold-start; OFF starts playback immediately and accepts brief silence (#98 Mode A).
• Punctuation Cadence — slider for inter-sentence pause (#90 / #109). On Kokoro, also drives silence_scale for within-sentence breaths (#196).
• Buffer (chunks) — render-ahead depth. Default 64. Experimental zone unlocks up to 3000 (#138).
• Engines (Tier 3) — 1–8 parallel VoxSherpa instances. Each adds ~50 MB RAM (Kokoro).
• Threads / engine (Tier 3) — thread-pool size per engine instance.
• PCM cache (v0.5.47–v0.5.49) — cache size cap + eviction priority. Streaming-tee writes synthesized PCM to disk while it plays; cache-hit playback skips re-synthesis on replay. Status icons in the chapter list show per-chapter cache state.

The producer thread is pinned to URGENT_AUDIO priority so it doesn't get descheduled mid-chapter.

AI

• Active provider — Claude / Teams / OpenAI / Vertex / Bedrock / Foundry / Ollama. See AI chat.
• Per-provider model chips — pick which model to use within each provider.
• Chat grounding — toggles for what context the chat sees (chapter title, current sentence, full chapter, full book so far).
• Cross-fiction memory (#217) — character/place/concept entities surface in per-book Notebook tab with manual edit.
• Function calling (#216) — invoke tools from chat ("Queue chapter 5", "Open Voice Library", "Mark this chapter read").
• Multi-modal image input (#215) — paste cover art / scene refs into chat.

Accessibility (v0.5.43)

• High contrast theme — higher-contrast Library Nocturne variant; auto-enables when TalkBack is active.
• Reduced motion — disable non-essential animations; auto-enables with system "Remove animations".
• Larger touch targets — bump tap-target minimums from 48dp to 64dp; auto-enables with Switch Access.
• Screen-reader pauses — slider (default 500 ms) for extra pause between sentences when TalkBack is active.
• Speak chapter numbers / titles — segmented control (Both / Numbers only / Titles only).
• Font scale override — multiplier on top of Android's system font scale.

AI sessions

• Sessions (#218) — review past chats, delete sessions, jump back into one.

Plugins

Brass-edged card grid iterating the @SourcePlugin registry — one card per backend. Search input, On / Off / All chips, tap-for-details modal. Three category sections: Fiction sources (25 in-tree), Audio streams (Radio), Voice bundles (v2 placeholder). Adding a new backend automatically surfaces a card. OCR, PDF, and LibriVox joined in v1.1.0; Telegram, Palace Project, Slack, and Matrix in v0.5.51 (the last three default OFF on fresh installs).

Pronunciation dictionary

Per-word phonetic overrides. Syncs across devices via InstantDB.

Account

• Royal Road — sign in / out via WebView. Unlocks Premium chapters and Follows sync.
• GitHub (#91) — OAuth Device Flow sign-in. Lifts the rate limit + unlocks My Repos / Starred / Gists tabs.
  • Enable private repos (#203) — opt-in, requires repo scope.
• AO3 (#426) — PR1 auth surface landed in v0.5.51; PR2 wires the WebView login.

Memory Palace

• Host — palace URL (e.g. http://palace.local:9080).
• API key — bearer token.
• Test connection — probes the palace and surfaces health.

Sync (InstantDB)

Cross-device sync of library, follows, positions, bookmarks, pronunciation, and encrypted secrets (#360 v0.5.12). Magical sign-in surface (v0.5.39+).

About

• Version + git hash + build time (realm-sigil at /version.json).
• Repo link — opens GitHub.
• License — GPL-3.0.

DataStore keys

All preferences persist in storyvox_settings.preferences_pb. Voice preferences live in a separate voices_settings.preferences_pb. Secrets (RR cookies, GitHub OAuth tokens, Azure key + region, AO3 cookies, bot tokens for Telegram/Slack/Matrix, Notion integration token) live in EncryptedSharedPreferences. InstantDB sync replicates library/follows/positions/bookmarks/pronunciation/secrets across devices. DataStore wipes only on uninstall + reinstall (or pm clear); upgrades preserve everything.

See also

• Reading & library — reader gestures: highlights & notes, find-in-chapter, shelves, library sort, add-by-URL
• Export & extras — audiobook (.m4b) export, EPUB export, the Now-Playing home-screen widget