Web Speech API Support

[Jeriff]

Please add support for https://webaudio.github.io/web-speech-api/

Chrome based browsers all support it.

Speech recognition in local browser.

[Jon]

Thanks for submitting an idea to the Mozilla Connect community! Your idea is now open to votes (aka kudos) and comments.

[flaqueman]

It needs very strict permission control since it will access the microphone. I would even argue that it needs to be opt-in so it's not abused

[boresox]

Why is the Retrosheet website always misaligned when I use firefox, but not for other browsers?

Thanks for any help on this,

Brad.  Long time Firefox Friend.   [email protected]

[Jeriff]

@flaqueman All Chrome-based browsers support that, that's 85% population. I don't think they have problems with it.

[flaqueman]

Of course it's fine for them... They're using Chrome, so clearly they don't care about privacy. Firefox users on the other hand do care about that and that's why it requires more privacy preserving features

[SugarPlacebo]

I would only be fine with something that runs locally on the device making any calls out and if it does not use up the memory and CPU. I personally have no use for such a feature even then but some may need something like I describe. 

[_wojtek]

Add support for WebSpeech API

supermemo.com allows using speech recognition for learning but firefox is not supported because... it lacks WebSpeech:

https://developer.mozilla.org/en-US/docs/Web/API/SpeechRecognition#browser_compatibility
https://caniuse.com/speech-recognition

There seems to be an meta-issue about sites broken by the lack of it:
https://bugzilla.mozilla.org/show_bug.cgi?id=1856507

BUT, it looks like there is currently no effort to have the support as the task to have it via DeesSpeech was closed: https://bugzilla.mozilla.org/show_bug.cgi?id=1474084

Maybe it could be done with other LLM model? Wisper?

[Jon]

Thanks for submitting an idea to the Mozilla Connect community! Your idea is now open to votes (aka kudos) and comments.

[Jon]

(Note: similar ideas have been merged into this thread)

[CougarBrenneman]

(Incidentally, I signed in before using what I thought was my password and I lost my entire comment, which was quite extensive, because no one informed me that my sign in was not complete. Please fix this.)

I currently have to tell people who come to my webpages that to implement the Web Speech API that they need to use Chrome. In other words, I tell them very specifically that they cannot use Firefox. I doubt that this is what you people want. I do agree that Chrome may not provide enough security. It's necessary to turn on permission to use the microphone to the page, and even though you do that once, that permission is not retained. In Windows, you go to that page, you have to click somewhere on the page to reactivate the permission. For android, there are side effects that no user will want to deal with, so for those purposes, I provided an alternative program-control methodology of using the Gamepad API to provide the user with a way of controlling the program on a Bluetooth battery-powered game controller.

However, I have some users that would prefer to have a way of using their voice to manage the android web app. The side effects of activating this form of program control in Chrome are that the phone produces a constant set of clicks and other noises that frankly dissuade me from ever using voice recognition while away from my laptop.

Please note that I am currently only using voice recognition for a limited set of alternatives to clicking buttons, and that I'm currently finding the limitations of this technology on Chrome to be annoying. There will be times in the future when I will wish to use voice recognition to allow people to fill in content into appropriate text boxes. Currently, these use cases are not in my immediate target arena, but they will be at some point.

In any case, I have been a fan of Firefox and Mozilla for decades, and I really don't like it that I have to advise people to use Chrome instead, but that's the case, right now. And no, adjusting permissions in about:config is not an acceptable workaround.

[hi_bye]

yes please add +1 

[Jon]

Goods news...

This is being worked on, and we are aiming to land it in Firefox 155 or 156 — stay tuned for updates and feel free to keep the conversation going. 

Thank you 🙌

[luis123456789]

@Jon  Given that communicated timeframe:

What considerations are being taken to protect user security and privacy given recent reveals about fingerprinting such as https://cybersecuritynews.com/aliexpress-webaudio-device-fingerprinting/ ?

At a minimum I would expect / need:

• WebAudio needs to be at least a per-site permission.
• Information processed by the API needs to be either available to extensions so they can randomize it, or be able to be exported at any point as a file so that users can diagnose.
• There needs to be some sort of onscreen indicator of when the feature is being used (similar to privacy beacons in Android).

[CougarBrenneman]

For my own purposes, the security issues make even the most generous permissions for speech recognition a little bit of a restriction problem. With Chrome, I can sit at my computer and do a guided meditation with my eyes closed by saying, "Next step," "Go back," or "Repeat that," using my Windows PC, and that is convenient enough, because it does not require me to open my eyes which would interfere with the meditation or trance experience. With an android, I have had to program a gamepad using the Gamepad API so that I can move forward in the meditation without looking at the screen on my phone. This distinction between the two systems is not a problem, because if I'm meditating in public with my android, it's convenient not to be forced to speak aloud to advance from step to step. I should also say that the security issues are stricter for android because even if I were trying to rely only on speech recognition, android requires me to push a button for every use of speech recognition. This destroys the advantage that speech recognition would give. I tend to use a headset, whether I'm on my phone or my computer, and as a solitary developer in an obscure usage category, there are some restrictions that I may just need to pass on to my users. It is often, however, unlikely to find users who are willing to use a headset for the kind of work I'm trying to do.

One other point is that I have active meditations that are done in the car while driving, and for these, using a gamepad to advance from step to step is appropriate, because stopping to look at the screen and push a button on the screen would be dangerous. It is going to be inconvenient for my users to be required to put a little square of Velcro on the dashboard — as I have done — to enable them to push buttons and advance the program without taking their eyes off the road. Requiring people to buy Velcro to get full use of my programs is a minor inconvenience, but there could be some people for whom it is too inconvenient.