Document novelty voices from Apple

[HadrienGardeur]

Apple preloads a number of novelty voices by default on macOS (multiple languages) and iOS/iPadOS (English only).

These novelty voices are more harmful than helpful for a read aloud feature and are unfortunately localized, which makes their detection more difficult than it should be.

With native APIs and Firefox, there might be ways of detecting them based on their package name, but that's not the case elsewhere.

In order to filter them out, we'll need to document these voices just like we do it for recommended voices. I'll use a separate JSON file specifically for that.

List of languages to support:

• Arabic
• Basque
• Bhojpuri
• Bulgarian
• Catalan
• Chinese
• Croatian
• Czech
• Danish
• Dutch
• English
• Finnish
• French
• Galician
• German
• Greek
• Hebrew
• Hindi
• Hungarian
• Indonesian
• Italian
• Japanese
• Kannada
• Korean
• Malay
• Marathi
• Norwegian
• Persian
• Polish
• Portuguese
• Romanian
• Russian
• Slovak
• Slovenian
• Spanish
• Swedish
• Tamil
• Telugu
• Thai
• Turkish
• Ukrainian
• Vietnamese

[HadrienGardeur]

After further investigation, it seems that these localized names are only used by Apple in its UI and not through the Web Speech API.

This is great news, since we can drop localized names and complete this task.

[HadrienGardeur]

"Bad news" (not the voice 😅) this is no longer the case on iOS 26, the voices are now translated everywhere.

But the "Good news" comes from voiceURI which contains a proper identifier in both Firefox and Safari now (but prefixed in Firefox). I've already documented this information in nativeID (which I might need to extend to all Apple voices at some point) which means that we can use it for Firefox/Safari in Readium Speech.

For Chrome, unfortunately the only option would be to document all translations.