Release v1.0.4: add multiformat read-aloud, examples, and demo video

Author: BoltzmannEntropy

.flutter-plugins-dependencies

@@ -8,6 +8,12 @@
   </p>
 </div>
 
+## Demo Video
+
+<video src="assets/mayari-video.mov" controls preload="metadata" playsinline width="960">
+  Your browser does not support the demo video. Download it from <a href="assets/mayari-video.mov">assets/mayari-video.mov</a>.
+</video>
+
 Mayari is an Apple-platform app for collecting, organizing, and exporting quotes from research PDFs. It combines a PDF reader, quote manager, native text-to-speech, and a markdown text reader in one workspace.
 
 ## Highlights
@@ -75,7 +81,7 @@ No HTTP servers. No API endpoints. Pure native communication via Flutter MethodC
 
 ### From DMG (Recommended)
 
-1. Download `Mayari-1.0.3.dmg` from [Releases](https://github.com/BoltzmannEntropy/Mayari/releases)
+1. Download `Mayari-1.0.4.dmg` from [Releases](https://github.com/BoltzmannEntropy/Mayari/releases)
 2. Open DMG and drag Mayari to Applications
 3. Right-click → Open (first launch only, for Gatekeeper)
 4. TTS model (~340MB) downloads automatically on first use

README.md

@@ -1,35 +1,35 @@
-# Mayari v1.0.3 Release Notes
+# Mayari v1.0.4 Release Notes
 
 **Release Date:** February 21, 2026
 
 ## Highlights
 
-- Website now features audio demos showcasing all 8 British TTS voices
-- Hero section includes sample audio player for quick voice preview
-- Voice Samples section with playable demos for Emma, Isabella, Alice, Lily, George, Fable, Lewis, and Daniel
+- Added EPUB and DOCX read-aloud support alongside PDF.
+- Added bundled Examples with ready-made sample documents and audiobooks for PDF, DOCX, and EPUB.
+- Added voice cards and language cards with language-based voice filtering.
+- Added repository demo video (`assets/mayari-video.mov`) and replaced the MayariWEB hero image with the demo video.
 
-## Previous Release (v1.0.2)
+## Read-Aloud and Audiobooks
 
-- Added explicit `Play`, `Pause`, and `Stop` controls on each audiobook card.
-- Kept queue-based audiobook generation with jobs in the left sidebar (`Jobs` tab).
-- Added model file location visibility in Settings (`Text-to-Speech` > `Model location`).
-- Removed hardcoded external-volume PDF defaults that triggered removable-volume access issues.
-- Added readable-file checks so inaccessible/stale PDF paths are filtered out safely.
-- Improved extraction error messages for denied file access.
+- Unified extraction pipeline now supports `.pdf`, `.docx`, and `.epub`.
+- Audiobook creation is available for extracted text across supported document formats.
+- Included non-placeholder example audiobook WAV files for all three sample formats.
 
-## Audiobook Workflow
+## Voice UX
 
-- `Create Audiobook` enqueues background jobs immediately.
-- Jobs are processed sequentially and persisted across restarts.
-- Failed jobs expose detailed error messages and can be retried/cancelled.
+- Voice cards now expose language metadata in the UI.
+- Language cards allow selecting one or multiple languages to filter visible voices.
+- Added queue test action to pre-generate language test audiobook jobs.
 
-## Stability
+## Validation
 
-- Startup no longer depends on hardcoded `/Volumes/...` paths.
-- Saved sources that are no longer readable are auto-pruned.
+- `dart analyze` reports no issues.
+- Full Flutter test suite passes.
+- Service tests validate document extraction (PDF/DOCX/EPUB), examples loading, and voice catalog behavior.
+- macOS debug and release builds complete successfully.
 
 ## Distribution
 
-- **Version:** `1.0.3+4`
-- **Primary Artifact:** `Mayari-1.0.3.dmg`
+- **Version:** `1.0.4+5`
+- **Primary Artifact:** `Mayari-1.0.4.dmg`
 - **Platform:** macOS 15.0+ (Apple Silicon)