Upload README.md

mobile-android/README.md

@@ -2,11 +2,7 @@
 
 Distributed mirror for the Android app's neural TTS assets. The seven zips here are downloaded by the app at runtime — first install of a neural engine pulls only what's needed (~74 MB for one Piper voice + shared data, or ~145 MB for the full Kokoro bundle), and the user can manage each model individually from inside the app.
 
-> Sibling layout:
->
-> - **`../extension/`** — Chrome extension models (raw HF format, published).
-> - **`../android/`** — local dev staging for sherpa-onnx upstream artifacts (not published).
-> - **`./mobile-android/` (you are here)** — the zips actually shipped to users (published to HF).
+> Sibling: **`../extension/`** holds the Chrome extension's neural models in raw HF format. Both subtrees share the same engine families (Piper VITS, Kokoro StyleTTS2) — only the on-disk packaging differs.
 
 ---
 
@@ -100,27 +96,21 @@ The native installer:
 
 ## Publishing workflow
 
-1. **Produce the zips** from `../android/`:
+1. **Produce the zips** from the local sherpa-onnx staging tree:
 
    ```sh
    pnpm --filter mobile produce:neural-zips
    ```
 
-   Writes to `packages/mobile/dist/neural-assets/`. The script reads sherpa-onnx upstream artifacts in `../android/{neural-28m,neural-82m}/` and emits the seven zips with the correct internal layouts.
+   Writes to `packages/mobile/dist/neural-assets/`. The script reads the locally-staged sherpa-onnx upstream artifacts (Piper voice ONNX files + Kokoro int8 v0.19 bundle + shared espeak-ng-data) and emits the seven zips with the correct internal layouts.
 
-2. **Copy into this folder** (for local convenience and snapshotting):
-
-   ```sh
-   cp packages/mobile/dist/neural-assets/*.zip "AI Neural Models/mobile-android/v1/"
-   ```
-
-3. **Upload to Hugging Face** — the user-facing primary mirror lives at:
+2. **Upload to Hugging Face** — the user-facing primary mirror lives at:
 
    <https://huggingface.co/Pixel-Labs/threadcast-neural-models/tree/main/mobile-android/v1>
 
    Drop all seven zips into that tree. HF preserves filenames as-is. No build steps, no metadata munging — they're just static assets behind a CDN.
 
-4. **(Optional) Mirror to a fallback host.** Same seven filenames at any HTTPS endpoint. Common picks:
+3. **(Optional) Mirror to a fallback host.** Same seven filenames at any HTTPS endpoint. Common picks:
    - Sibling HF repo (`Pixel-Labs/threadcast-neural-models-mirror/v1/...`)
    - GitHub Release with `gh release upload v1 dist/neural-assets/*.zip`
    - Private CDN (R2, S3, etc.)