Frame 69 of 107 · windows-narrator-voice

Per-language narrator voice pickers showing an automatic choice and a language with no installed voice

1280 × 900 px · 127,208 bytes · batch windows-narrator-voice

呢一頁只講一張相：邊個 scene 出，邊個 batch 收，尺寸、指令、憑證全部照抄記錄，冇記錄就照直講冇。

Settings then Sound showing both narrator voice pickers, English reporting that a voice was chosen automatically and Cantonese reporting that this computer has no voice installed that can read the language — Per-language narrator voice pickers showing an automatic choice and a language with no installed voice

What the sources record

Asset file: docs/assets/screenshots/narrator-voice-pickers.png
Pixel dimensions: 1,280 × 900 px, read from the PNG IHDR header
File size: 127,208 bytes on disk
PNG encoding: 8-bit truecolour (IHDR colour type 2)
Gallery output id: narrator-voice-pickers
Scene: narrator-voice-pickers
Capture batch: windows-narrator-voice
Platform: windows-headless
Feature Gallery section: Search, history, accounts, and automation
Guided workflow caption: Per-language narrator voice pickers showing an automatic choice and a language with no installed voice
Publication status: Published target in the current Windows guided gallery

Alternative text

Settings then Sound showing both narrator voice pickers, English reporting that a voice was chosen automatically and Cantonese reporting that this computer has no voice installed that can read the language

The alternative text above is the Feature Gallery's own Markdown alt text for this frame, so the image describes itself identically in the wiki and here.

The interaction the harness performs

Open Settings from the application menu, select Sound, scroll both narrator voice pickers into view and drop focus. The capture host has English voices and no Cantonese one, so the frame shows the automatic and no-voice-available states together.

Fixture the capture batch requires

No repository. The capture host had English voices installed and no Cantonese one, which is why the frame shows both honest states at once rather than one of them twice.

Privacy gate the capture must pass

A throwaway user-data directory and a fake HOME for Git. Installed voice names are platform-provided and carry no user data.

Regenerating this capture

Run these commands in order, exactly as the capture batch records them. Placeholders in angle brackets are the verifier's own run root, fixture path and CDP port.

Step 1 of 2
```
yarn compile:prod
```

Step 2 of 2

node script/capture-app.js --out=<owned-run-root>
arrator-voice-pickers.png --size=1280x900 --step=menu:show-preferences --step=wait:1500 --step=click-text:Sound --step=wait:900 --step=scroll-to:.sound-voice-picker --step=wait:700 --step=blur

Dated acceptance receipt

No dated receipt under docs/verification/ names narrator-voice-pickers.png. The frame is published without its own dated acceptance document; the build, interaction and privacy evidence for its batch is recorded in HANDOFF.md and in the Feature Gallery's refresh notes instead.