Why isn't TTS and SST FOSS and shipped in the browser circa 2024?

[guest271314]

I was browsing through the configuration folder of Chromium 122 the other day and came across the screen_ai folder. It's not clear what the code (extension?) does from the manifest.json. 

{
  "manifest_version": 2,
  "name": "Chrome Screen AI",
  "version": "122.1"
}

There's a shared library file libchromescreenai.so that is 288 MB alone. 

That got me to thinking about how an "AI" program with a ~300 MB shared library file wound up in Chromium source code before speech synthesis (PATTS, et al) and speech recognition technologies - which still make network requests to remote servers for Web Speech API implementation - when Google voices are used for SpeechSynthesisItterance, and all of the time for webkitSpeechRecognition. Can't be file size restraints given the above.

Why isn't TTS and SST FOSS and shipped in the browser circa 2024?