fix: downgrade Flatpak runtime to GNOME 48 for wheel compatibility

.github/workflows/build.yml

@@ -18,34 +18,25 @@ jobs:
       - name: Install test tools
         run: python -m pip install --upgrade pytest ruff
       - name: Compile source
-        run: python -m compileall -q voice2text voice2text_ai.py qt_app.py
+        run: python -m compileall -q voice2text voice2text_ai.py
       - name: Lint
-        run: ruff check voice2text tests voice2text_ai.py qt_app.py
+        run: ruff check voice2text tests voice2text_ai.py
       - name: Unit tests
         run: python -m pytest -q
 
   flatpak:
     needs: test
     runs-on: ubuntu-latest
     container:
-      image: ghcr.io/flathub-infra/flatpak-github-actions:gnome-50
+      image: ghcr.io/flathub-infra/flatpak-github-actions:gnome-48
       options: --privileged
     steps:
       - uses: actions/checkout@v4
       - name: Generate pinned Python sources
         shell: bash
         run: |
           git clone --depth=1 https://github.com/flatpak/flatpak-builder-tools.git /tmp/flatpak-builder-tools
-          python3 /tmp/flatpak-builder-tools/pip/flatpak-pip-generator \
-            --requirements-file=packaging/flatpak/requirements.txt \
-
-            --runtime org.gnome.Sdk//50 \
-
-            --prefer-wheels=ctranslate2,onnxruntime,tokenizers,av,numpy,pyyaml,protobuf,hf-xet \
-
-            --wheel-arches=x86_64 \
-
-            --output=python3-requirements-flatpak
+          python3 /tmp/flatpak-builder-tools/pip/flatpak-pip-generator --requirements-file=packaging/flatpak/requirements.txt --runtime org.gnome.Sdk//48 --prefer-wheels=ctranslate2,onnxruntime,tokenizers,av,numpy,pyyaml,protobuf,aiohttp --wheel-arches=x86_64 --output=python3-requirements-flatpak
       - uses: flatpak/flatpak-github-actions/flatpak-builder@v6
         with:
           bundle: Voice2Text-AI.flatpak

README.md

@@ -1,12 +1,99 @@
-# Voice2Text AI microphone dropdown readability fix
+# Voice2Text AI
 
-This overlay updates the existing `release/v0.4.0-native-linux` branch.
+Fast native Linux dictation with local AI. Record from your microphone, have
+speech transcribed on-device with Faster Whisper, prompt a local Ollama model,
+and hear the answer spoken back with a natural neural voice and an offline
+eSpeak NG fallback.
 
-Changes:
+Built with GTK 4 and libadwaita, powered by GStreamer, and distributed as a
+Flatpak. Works on Wayland and X11 with PipeWire or PulseAudio.
 
-- Widens the Preferences dialog.
-- Shows the complete selected microphone name as the row subtitle and tooltip.
-- Uses a wider selected-value label with middle ellipsis only when unavoidable.
-- Wraps every microphone source name in the dropdown list, so long PipeWire,
-  ALSA, USB, monitor, and hardware source names can be read in full.
-- Does not alter dictation, rendering, TTS, Flatpak dependencies, or settings.
+## Features
+
+- **On-device dictation** — microphone audio is transcribed locally by Faster
+  Whisper (nothing leaves your machine).
+- **Local AI prompts** — send the transcript to any Ollama model you have
+  pulled, with streaming responses.
+- **Natural speech output** — Edge TTS voices stream immediately; if the
+  network voice is unavailable, eSpeak NG speaks offline automatically.
+- **Adaptive segmentation** — speech is split at pauses, so dictation flows
+  naturally while transcribing in the background.
+- **Coordinated appearance** — follows the system light/dark setting, with an
+  explicit override in Preferences.
+
+## Install (Flatpak)
+
+The easiest way is the Flatpak from the [releases](https://github.com/crhy/Voice2Text-AI/releases):
+
+```bash
+flatpak install --user Voice2Text-AI.flatpak
+flatpak run io.github.crhy.voice2textai
+```
+
+For local AI, install [Ollama](https://ollama.com/) and pull a model, for example:
+
+```bash
+ollama pull llama3.1:8b
+```
+
+The Whisper model downloads on first launch (the `base` model is the default;
+smaller models use less memory and start faster).
+
+## Run from source
+
+Requires Python 3.11+, GTK 4, libadwaita, and GStreamer with the Python
+bindings. Install the Python dependencies and launch:
+
+```bash
+python3 -m venv .venv
+.venv/bin/pip install -r requirements.txt
+.venv/bin/python voice2text_ai.py
+```
+
+`voice2text_ai.py` is a thin launcher over the `voice2text` package; the
+PyPI entry point is `voice2text-ai`.
+
+## Keyboard shortcuts
+
+| Shortcut            | Action                        |
+| ------------------- | ----------------------------- |
+| `Ctrl+R`            | Start or stop dictation       |
+| `Ctrl+Enter`        | Ask AI                        |
+| `Ctrl+Shift+C`      | Copy transcript               |
+| `Ctrl+L`            | Clear                         |
+| `Ctrl+,`            | Preferences                   |
+| `Ctrl+Q`            | Quit                          |
+
+## Building the Flatpak
+
+The GitHub Actions workflow generates pinned Python dependencies, builds the
+Flatpak bundle, and attaches it to a draft release for every `v*` tag. To build
+locally, generate the pinned module and run flatpak-builder:
+
+```bash
+git clone https://github.com/flatpak/flatpak-builder-tools.git /tmp/flatpak-builder-tools
+python3 /tmp/flatpak-builder-tools/pip/flatpak-pip-generator \
+  --requirements-file=packaging/flatpak/requirements.txt \
+  --runtime org.gnome.Sdk//48 \
+  --prefer-wheels=ctranslate2,onnxruntime,tokenizers,av,numpy,pyyaml,protobuf \
+  --wheel-arches=x86_64 \
+  --output=python3-requirements-flatpak
+flatpak-builder --user --install --force-clean build-dir io.github.crhy.voice2textai.yml
+```
+
+## Configuration
+
+Settings are stored in `$XDG_CONFIG_HOME/voice2text-ai/config.json` and edited
+from the Preferences dialog.
+
+## Acknowledgements
+
+- [Faster Whisper](https://github.com/SYSTRAN/faster-whisper) for on-device transcription
+- [Ollama](https://ollama.com/) for local language models
+- [Edge TTS](https://github.com/rany2/edge-tts) for natural voices
+- [eSpeak NG](https://github.com/espeak-ng/espeak-ng) for offline speech
+- GTK 4, libadwaita, and GStreamer
+
+## License
+
+MIT — see [LICENSE](LICENSE).

io.github.crhy.voice2textai.metainfo.xml

@@ -23,7 +23,7 @@
   </screenshots>
   <content_rating type="oars-1.1" />
   <releases>
-    <release version="0.4.0" date="2026-07-29">
+    <release version="0.4.0" date="2026-08-02">
       <description>
         <p>Rebuilt around GTK 4, libadwaita, and GStreamer with a modular, lower-overhead runtime.</p>
       </description>

io.github.crhy.voice2textai.yml

@@ -1,6 +1,6 @@
 id: io.github.crhy.voice2textai
 runtime: org.gnome.Platform
-runtime-version: '50'
+runtime-version: '48'
 sdk: org.gnome.Sdk
 command: voice2text-ai