We are decommissioning the Mozilla Speech Proxy Server on February 28, 2021. This server helped power the Voice Fill and Firefox Voice Beta extensions, and as a result of the server decommissioning, we will be retiring these extensions.
We plan to update these add-ons to notify users about the retirement timeline. The extensions will uninstall themselves on February 19, 2021. If you would like to learn more about the retirement of these extensions, please visit this support article.
Thank you to everyone who participated in and contributed to Mozilla’s projects to bring voice data and deep learning algorithms available to the open source world. The source code for the Voice Fill and Firefox Voice Beta extensions will be freely available and distributed under the Mozilla Public License v 2.0.
Hi @inyazserg. I think you are mentioning Firefox Voice maybe? Voice Fill does not listen to your microphone all the time looking for a keyword to trigger the speech recognition in an Alexa’s style and does not open any websites or provide answers. It’s simply a voice search functionality.
Firefox Voice yes, has a more voice assistant-ish experience as you described.
You are right, I meant Firefox voice, the extension which represents the first step to making a voice-controlled browser. You can minimize windows with your voice, for example. If you want to close the current tab, you can just say “hey Firefox, close the window”.
Voice Fill (extension) cannot close your tabs, but it is an effective tool for typing with your voice into Google search field, unless your walls are thin and you are shy.
I am disabled, can’t type, need voice search, and ability to turn mic off/on by 1 click, like Voice Fill. Need Voice Fill for any fill-able field/form.
Sorry to see this end so soon. I wish I used the voice features more… I honestly was very excited when they popped up in FF the first time. But I honestly forgot about it and only used it once or twice… I wish I had remembered to use it more.
It’s a shame to see this enabling feature discontinued so soo with next to no advertising/promoting of its beneficial features. I will have to go where such a feature is enabled
If you take Voice Fill you must replace it with something better, or fix the microphone setting in Firefox that wont allow microphone on your prefer web sites.
Does this mean that Mozilla is abandoning voice-based interfaces altogether and leaving the field to privacy-invasive actors like Google and Alexa? Please say it ain’t so! If Mozilla has an alternate roadmap w.r. to voice initiatives and/or is aware of any other privacy-respecting, open-source initiatives in this field, please advise…
@caitlin Hello. Are there current alternatives to these extensions currently available? They were really helpful to me, particularly on Linux, where tools like Dragon NaturallySpeaking and Windows speech recognition are not available.
Hey, Mozilla. Here’s a thought. Is there any way that you could make an offline version of the Firefox Voice and Firefox Voice Fill extensions? In other words, is there any way that you could possibly modify the extensions so that they can be powered by users’ computers instead of the servers? Maybe you could combine them into one extension, and modify it to be able to run locally on a user’s computer? Thanks. God bless. Jesus Christ loves you - He always has and always will, no matter what! You’re priceless and you matter!
Unfortunately, this is not feasible quality wise, given that even with popular projects like Common Voice and Deep Speech, we never got enough representation to train large speech-to-text models as good as the online ones we used for Voice Fill and Firefox Voice to cover the whole diversity of our user base.
Thanks for your enthusiasm though and for being a user through all this time.
It’s a real shame that this is being retired but I understand having to reprioritize. I was hoping we could add mozilla deep speech is a backend to the dragonfly project but without community support/Mozilla it seems unlikely.
I work with products such as Caster which builds off of Dragonfly framework. The goal of these frameworks is to allow people to control their compute environment by Voice. This allows people disabilities interact in a way of the computer that they would not otherwise be able to or for developers to augment the workflow. We continue to push for developing open source solutions for voice recognition stack.
Part of an open source solution Kaldi https://github.com/daanzu/kaldi-active-grammar where we are developing models for voice recognition commands and dictation. The pre-trained model general English Kaldi nnet3 chain mode is trained on ~3000 hours of open audio. Perhaps that’ll be helpful those who rely on Firefox voice.
