Configure static build and isolation headers

README.md

@@ -1,10 +1,34 @@
 ---
 title: Phonon Reader
-emoji: 📉
-colorFrom: gray
+emoji: 🔊
+colorFrom: yellow
 colorTo: red
 sdk: static
+app_file: dist/index.html
+app_build_command: npm install && npm run build
 pinned: false
+short_description: Paste text, listen. On-device TTS, nothing uploaded.
+models:
+  - gradium/phonon-31m-en
+custom_headers:
+  cross-origin-embedder-policy: require-corp
+  cross-origin-opener-policy: same-origin
+  cross-origin-resource-policy: cross-origin
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Phonon Reader
+
+A small installable web app: paste an article or any text and listen to it.
+Speech is generated **in your browser** with [Phonon 31m](https://huggingface.co/gradium/phonon-31m-en)
+via [`phonon-tts`](https://www.npmjs.com/package/phonon-tts) (WebAssembly). Your text is never uploaded.
+
+- First run downloads the model (~65 MB) from the Hub; after that it works offline.
+- Install it: open the Space's direct URL in Chrome on Android and choose **Install app**.
+  It then shows up in the Android share sheet, so you can share selected text into it.
+- Text is split into sentence chunks, generated a few chunks ahead of playback, and highlighted as it's read.
+- Voices come from [`gradium/phonon-voices`](https://huggingface.co/gradium/phonon-voices).
+
+Network requests: the app files from this Space, and model/voice files from huggingface.co. Nothing else.
+
+No bundler: `npm run build` copies the app and the `phonon-tts` package into `dist/` and writes the
+service worker's precache list. `npm run serve` serves `dist/` locally with the isolation headers.