Open Bug 1244237 Opened 8 years ago Updated 11 months ago

enable Web Speech API for desktop

Categories

(Core :: Web Speech, enhancement)

enhancement

Tracking

()

People

(Reporter: dietrich, Unassigned)

References

(Blocks 2 open bugs)

Details

(Keywords: dev-doc-needed)

Basic implementation is complete, and enabled in Firefox OS. Enabling in desktop is just a build flag. API is already shipping in Chrome and Opera.

Let's enable this on desktop.

If for some reason we're not ready to enable in release channels, let's at least make it available to developers somehow (Nightly, Dev Edition, about:config, etc).
Andrew, how do we get started with this process?
Flags: needinfo?(overholt)
Hm, looks like maybe already in about:config via media.webspeech.recognition.enable.

What's blocking turning that on by default?
Blocks: 1244242
Blocks: 1244243
Is this issue for Web Speech recognition API?  Web Speech API has 2 parts (synthesis and recognition).
This is for both parts.
Blocks: 1244460
Depends on: TTS_for_firefox
(In reply to Dietrich Ayala (:dietrich) from comment #1)
> Andrew, how do we get started with this process?

Sending an intent to ship email is a decent way to kickstart the conversation: https://wiki.mozilla.org/WebAPI/ExposureGuidelines#Email_templates
Flags: needinfo?(overholt)
MDN article: https://developer.mozilla.org/docs/Web/API/Web_Speech_API/Using_the_Web_Speech_API

For speech recognition MDN says:

"Firefox desktop and mobile support it in Gecko 44+, without prefixes, and it can be turned on by flipping the media.webspeech.recognition.enable flag to true in about:config. The permissions settings/UI haven't yet been sorted out however, so permission can't be granted to use it by the user, so it can't be used. This will be fixed soon."

I don't know what bug it is which will be fixing it soon. ?

For speech synthesis MDN says:

"Firefox desktop and mobile support it in Gecko 42+ (Windows)/44+, without prefixes, and it can be turned on by flipping the media.webspeech.synth.enabled flag to true in about:config."

That sounds ready to go.
Depends on: 1248897
(In reply to Jefferson from comment #6)
> MDN article:
> https://developer.mozilla.org/docs/Web/API/Web_Speech_API/
> Using_the_Web_Speech_API
> 
> For speech recognition MDN says:
> 
> "Firefox desktop and mobile support it in Gecko 44+, without prefixes, and
> it can be turned on by flipping the media.webspeech.recognition.enable flag
> to true in about:config. The permissions settings/UI haven't yet been sorted
> out however, so permission can't be granted to use it by the user, so it
> can't be used. This will be fixed soon."
> 
> I don't know what bug it is which will be fixing it soon. ?

I also couldn't fine one. Though it looks like the missing thing is that the SpeechRecognition API is not exposed to website content yet. So, I've created bug 1248897.

Sebastian
No longer depends on: 1248897
Depends on: 1268633
I think the issue is that the recognition part (speech to text) hasn't been implemented yet. If you look into the speechRecognition.cpp file, you can just return true everywhere it checks for a permission and websites will get access to the api. However things still don't work because of the lack of recognition part (at least that's what I feel looking at the source). It would be great if anybody can point towards the team working on this, would love to help.
One needs to build Firefox with the flags turned on to include the only current implementation available (pocketsphinx + english). If you guys are willing to move forward with it, I can help you to set and have it running.
(In reply to André Natal from comment #9)
> One needs to build Firefox with the flags turned on to include the only
> current implementation available (pocketsphinx + english). If you guys are
> willing to move forward with it, I can help you to set and have it running.

Interested.

Just as a point for voice in the browser (not necessarily the API itself), I would love to be able to dictate long-form texts like blog posts to the browser (and then probably proof-read and improve them).

Type: defect → enhancement

Release Note Request (optional, but appreciated)
[Why is this notable]: Cool new feature
[Affects Firefox for Android]: No
[Suggested wording]: Web Speech API enabled (nightly only)
[Links (documentation, blog post, etc)]: dunno?

relnote-firefox: --- → ?

Removing this request to get it off the relnote radar for now. Please re-nominate if/when the status changes.

relnote-firefox: ? → ---
Severity: normal → S3

In our company, we successfully integrated vosk open source project for offline speech recognition on desktop. It is probably possible to implement it on Firefox too. It gives pretty accurate results, as long as the language model is being downloaded for the user language.

Would it be better than the current solution?

This is also affecting Axis Bank's site, which seems to presume the webkitSpeechRecognition API is always available (as found in https://webcompat.com/issues/117770) and fails to show a credit card details page otherwise.

You need to log in before you can comment on or make changes to this bug.