Cozy Read Aloud Mobile 0.3.10

Cozy Read Aloud Mobile 0.3.10

Formerly Note and PDF Voice Reader for Mobile. The plugin ID, settings and update path are unchanged; no reinstall is required.

• Clarify the short description with read aloud and voice terminology.
• Add the former name to the English and Chinese introductions and community-description source.

Documentation and metadata only. No changes to playback or synthesis.

Cozy Read Aloud Mobile 0.3.9

Cozy Read Aloud Mobile 0.3.9

• Rename the mobile edition to Cozy Read Aloud Mobile.
• Update the reader title, documentation, community-description source and service application label.
• Keep the plugin ID, repository, saved settings and update path unchanged. No reinstall is required.

This release does not change audio synthesis or document processing. Existing license attribution and historical release notes are preserved.

0.3.8

0.3.8

• Simplify setup into System speech, Online speech and Self-hosted service, preserving existing provider settings.

• Add a fixed-sample voice test with Stop test and bounded error guidance. It does not read notes or interrupt reading/export; online tests may incur charges.

• Collapse advanced options and improve narrow-screen settings layout.

• Remove duplicate dividers at the top of engine and academic settings.

• Add Read aloud from here and Read selected text to Obsidian's Markdown editor context menu. Availability on mobile depends on Obsidian exposing that menu; this does not replace the native selection popup. PDF, HTML and reading views retain toolbar actions.

• Independently toggle editor menus and the optional Editing Toolbar speaker in Playback settings. Choose selected text (default) or selection-to-end. Changing these preferences does not stop playback.

Validation: 149 tests and mobile compatibility checks passed. Shared desktop/mobile settings passed 24 browser layout checks. These changes have not yet been verified on a physical iPhone or iPad.

0.3.7 - Optional prefetch and session audio cache

0.3.7 - Optional prefetch and session audio cache

• Add an optional Prepare next audio part setting, off by default. When enabled, online synthesis is limited to two simultaneous operations and one future audio part.
• Reuse in-flight requests and recently prepared audio in a bounded, memory-only session cache. Cached revisits avoid unnecessary synthesis.
• Prioritize the latest queued seek target and discard stale results after stop or navigation. Pause prevents queued requests from starting.
• Clear cached audio on stop, new document, relevant configuration changes, or plugin unload. Exported files and saved reading positions are not cache files.
• Add an on-demand command to copy numeric waiting-time statistics without note text or credentials. No automatic telemetry.

Validation: 140 tests and the mobile compatibility boundary check passed. This release has not yet received new iPhone/iPad device validation. Prefetch may synthesize audio that is not ultimately played, and already-sent requests may still be billed. Provider quotas apply; no zero-delay playback guarantee is made.

0.3.6 - Model-specific price references

0.3.6 - Model-specific price references

• OpenRouter settings show dated price references for the four ElevenLabs presets and verified MAI models, with per-million and per-10,000-character costs where applicable.

• The ElevenLabs launch offer has an explicit deadline. Settings rendered after it ends show only the last checked list price and a reminder to verify current pricing.

• Unknown models show no numeric estimate. The official-pricing button follows the selected model; price hints make no background requests and never send note text or keys.

• English and Chinese copy distinguishes reference prices from actual charges and warns about repeat synthesis, fees and taxes.

• Playback, privacy consent and ZDR routing are unchanged. No paid synthesis or new real-device iOS test was performed for this documentation/settings update.

0.3.5 - ElevenLabs options and everyday pronunciation examples

0.3.5 - ElevenLabs options and everyday pronunciation examples

• Add four ElevenLabs models and six voice presets through OpenRouter without changing existing defaults.
• Offer optional, default-off continuity controls only for supported models, with mandatory ZDR and bounded context from the selected reading range.
• Add local term pronunciation rules for every engine and audio export. Examples use familiar abbreviations: AI = artificial intelligence in English and AI = 人工智能 in Chinese.
• Preserve original document text and reading positions while keeping expanded requests within engine limits.
• Include the recent interval-formula improvements and normal-speed ElevenLabs synthesis with player speed controls.

No paid listening tests or new iOS/iPadOS device tests were performed. Continuity remains experimental; improved prosody is not guaranteed. No cloud dictionary is uploaded. Online engines receive substituted speech text, not the rule list.

使用提示

选择受支持的 ElevenLabs 模型后，可在高级选项中开启分段衔接实验功能。通用的本地术语读法位于“学术阅读”，例如 AI = 人工智能，将缩写展开为完整读法，不修改笔记原文。

0.3.3 - Short formula recognition

Fix short LaTeX variables incorrectly omitted because of font commands, redundant atom groups, or layout spacing. Uses core v0.2.1. Explicit skip settings and safeguards for complex or unsupported formulas are unchanged. Includes 117 passing tests and mobile boundary checks; no new iOS device test is claimed.

0.3.2 - Scrollable mobile player and data controls

0.3.2 - Mobile panel spacing and data controls

• Give the expanded player a bounded height and vertical touch scrolling, and reserve bottom space so the final export button can move above floating native navigation. Tablets use a smaller safe-area allowance.
• Add separate Privacy controls for clearing the in-memory outline and saved reading positions. Position deletion requires confirmation; playback, documents, keys and exported audio are retained.
• Document audio lifecycle and WAV export prerequisites: an online engine and an existing reading queue. Export synthesizes the queue again and may incur provider charges.

Automated checks pass. The bottom-spacing change still needs verification on the reported iPhone; no claim of complete iOS/iPad/Android device validation is made.

0.3.1 - Sentence-first reading and document outlines

0.3.1 - Sentence-first reading and document outlines

• Prefer complete sentences over filling the current audio chunk. Only an oversized sentence is split; MiMo retains the 200-character cap. Preserve text and avoid splitting surrogate pairs.
• Conservatively join PDF cross-page continuations and retain the starting page metadata.
• Add Markdown and local HTML heading outlines, section reading and cached extraction. HTML uses h1-h6 headings and requires HTML Reader for source navigation.
• Click an outline title to locate the source without starting speech. Use the adjacent range button to select Read section or Read from section.
• Locate PDF headings with available viewer coordinates; otherwise open the corresponding page with an explicit fallback notice. Keep the outline open when HTML navigation cannot be resolved safely.
• Update bilingual documentation and community copy. Desktop and shared core repositories are unchanged.

Automated tests, browser-only API checks and production build pass. These additions still require broader iOS/iPad/Android real-device testing. The previously user-confirmed basic iPhone workflow does not validate every new navigation feature or PDF layout.

0.3.0 - PDF outlines, HTML reading and WAV export

0.3.0 - Mobile PDF Sections, HTML and Audio Export

Public name: Note and PDF Voice Reader for Mobile. Plugin ID remains note-reader-mobile; existing settings and installation paths are unchanged.

• Add a searchable, dismissible PDF outline panel: use built-in bookmarks first, otherwise infer headings and preserve numbering. Read a section (including children) or continue to the end.
• Cache one PDF outline in memory, invalidate when the source/header setting changes, and provide manual refresh. Do not write back to PDFs.
• Add conservative footnote separation, including two-column page-bottom notes and first-page correspondence blocks. Default to body-only, with original-order and footnotes-only options; add a toolbar footnote scope.
• Add local HTML/HTM reading with HTML Reader, including selected text, exact rendered selected position, full file and resume. Source parsing never executes scripts or downloads page resources.
• Export an online-engine reading range to mono 16-bit WAV after explicit confirmation. Save under Note Reader Audio/ without overwriting; allow cancellation and enforce 30,000-character/32-MiB limits. Export makes new requests and may incur charges; system speech export is unavailable.
• Update English/Chinese documentation, community description, privacy/feedback guidance and device checklist.

Validation: automated regression tests, production build, mobile API boundary and public-file privacy checks. No paid speech requests or private documents were used in automated tests. GitHub checks and asset hashes are verified during publication.

The user confirmed the prior basic iPhone workflow: full-file reading, selected-position reading, pause/resume, Stop, speed and an unobstructed toolbar. This is not a claim that the new features or every PDF/iPad/Android combination have passed device tests. Report reproducible PDF issues with redacted reader/PDF screenshots; share files only when permitted.

This is an early mobile prerelease prepared for the user-requested community submission. Community listing/review status is independent of this GitHub release. Desktop and shared-core repositories are unchanged.