Updates to Web Speech API data

[chrisdavidmills]

Summary

This PR makes several changes to the Web Speech API data:

• On-device speech recognition (https://chromestatus.com/feature/6090916291674112). The ChromeStatus entry says 139, but the engineer said it has been slipped to a 142 release because of some bugs. It is also only supported on Chrome desktop for now, not Android/WebView
  • The on-device-speech-recognition Permissions-Policy directive - newly added
  • SpeechRecognition.processLocally - already in, but updated Chrome support from 139 to 142 and set Android support to false
  • SpeechRecognition.available() - already in, but updated Chrome support from 139 to 142 and set Android support to false
  • SpeechRecognition.install() - already in, but updated Chrome support from 139 to 142 and set Android support to false
• Web speech API contextual biasing (https://chromestatus.com/feature/5225615177023488). ChromeStatus says 142, which is correct according to the engineer
  • SpeechRecognition.phrases - newly added
  • SpeechRecognitionPhrase interface - newly added
• Speech recognition from a MediaStreamTrack (https://chromestatus.com/feature/5178378197139456). ChromeStatus says 135, which is correct according to the engineer
  • SpeechRecognition.start() audioTrack parameter - newly added
• Removes the following items (as far as the engineer can tell, these were removed from the Chromium implementation and the spec about 5 years ago):
  • SpeechRecognitionEvent.emma
  • SpeechRecognitionEvent.interpretation
• Changes the standards-track settings for SpeechGrammar and SpeechGrammarList
  • All documented features are still in the spec except for the SpeechGrammar() constructor, so that one should be "standard_track": false but the others should all be "standard_track": true. All of them should be "deprecated": true.
  • I've also removed the weird Firefox support notes that say Speech Synthesis is supported in Firefox (doesn't make sense to say it in Speech Recognition features)
  • I've also made the Firefox support data consistent between these two interfaces (it really doesn't make sense for Firefox to have implemented one of these but not the other, so I think they were implemented at the same time)

Test results and supporting details

Related issues

[github-actions]

Tip: Review these changes grouped by change (recommended for most PRs), or grouped by feature (for large PRs).

[caugner]

Some preliminary remarks, didn't check everything yet.

Overall, I find this PR hard to review. Could you please consider splitting it up? I think it would make sense to separate:

• the new permission
• the new phrase-related features
• the Firefox fixes
• cleaning up the SpeechRecognitionEvent

[chrisdavidmills]

Overall, I find this PR hard to review. Could you please consider splitting it up? I think it would make sense to separate:

• the new permission
• the new phrase-related features
• the Firefox fixes
• cleaning up the SpeechRecognitionEvent

OK, I mean, it's not even that big a set of changes, and they are all related, more or less. Can you not just concentrate on one of those bullets per day until they are all done? I'm happy to wait a few days.

[caugner]

FWIW, Firefox also supports the webkit prefix for these interfaces and their constructors:

[chrisdavidmills]

So you probably want me to update the support data for each one of these interfaces and their constructors to something like

"firefox": [
  {
    "version_added": "142",
    "flags": [
      {
        "type": "preference",
        "name": "media.webspeech.recognition.enable",
        "value_to_set": "true"
      }
    ]
  },
  {
    "prefix": "webkit",
    "version_added": "33"
  }
]

But, which browser version?

[caugner]

Yes, only the interfaces and the constructors, with the same version and flags:

[
  {
    "version_added": "142",
    "flags": [
      {
        "type": "preference",
        "name": "media.webspeech.recognition.enable",
        "value_to_set": "true"
      }
    ]
  },
  {
    "version_added": "142",
    "flags": [
      {
        "type": "preference",
        "name": "media.webspeech.recognition.enable",
        "value_to_set": "true"
      }
    ],
    "prefix": "webkit"
  }
]

[chrisdavidmills]

OK, done

[caugner]

As mentioned above, Firefox also supports SpeechRecognition behind the pref, confirmed with https://mdn-bcd-collector.gooborg.com/tests/api/SpeechRecognition in Firefox 143.

[chrisdavidmills]

OK, I've added Firefox 143 support for SpecchRecognition and all relevant subfeatures.

[caugner]

Addendum: Firefox also supports SpeechGrammar only since 142 behind the pref.