Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Firefox Voice

voice.mozilla.org

Post reply on HN
Print view View original

121–130 of 166 posts

Re: Firefox Voice

#121

Post by ptx »

fuzxi wrote: post #101
cpeterso wrote: post #66

Earlier quoted context omitted.

The new Firefox for Android (aka "Fenix") supports extensions that just uBlock Origin. It supports the following at this time, with more coming soon: uBlock Origin Dark Reader HTTPS Everywhere Privacy Badger NoScript Security Suite Decentraleyes Search by Image YouTube High Definition Privacy Possum https://www.androidpolice.com/2020/07/28/mozillas-next-gen-f...

That's frankly a pathetic list when you consider not only the sum total of available extensions, but the list of Recommended Extensions alone. There's a Github issue for this with one particularly interesting reply: [1] >Let it be put this way. I have extensions which I can already test while tethered using web-ext. They work. I know they work. But for some reason, Mozilla seems adamantly opposed to letting me test t…

> Sideloading unapproved extensions will probably never be allowed because of security and performance implications

Very wise. They should apply those same principles to the Rust project - the code I write might very well have issues with security and performance, so it would be better if the compiler submitted it to Mozilla and awaited their approval before allowing me (or potentially not allowing me) to produce an executable.

Re: Firefox Voice

#122

Post by Cybiote »

panpanna wrote: post #111
AsyncAwait wrote: post #107

I see a lot of skeptical voices here, (somewhat warranted, given it's a voice assistant technology), but the fact remains that if we want open, on-device voice recognition, we'll have to do the work and donate sample data. This extension is trying to provide some useful functionality in the hopes that Mozilla gets more data for https://commonvoice.mozilla.org I'd at least consider recording your voice, especially if…

> if we want open, on-device voice recognition, we'll have to do the work and donate sample data. Fair enough, but is any stopping Mozilla from _also_ selling your voice data to third parties including advertisers and commercial ML interests? (Asking this because Firefox has started sharing our data with leanplum)

It still remains a risk well worth taking. With Mozilla, it is merely uncertain but for just about every other company, it is all but guaranteed. You can plainly see this philosophy in the industry's naming sense, where super-computers of yester-year are relegated to "edge" roles.

As of today, the open source and free software equivalents to machine learning and AI products are sorely lacking when compared to commercial offerings. Whether it is open-ended speech to text with good ergonomics, text to speech, intent recognition, speaker recognition, OCR for text, OCR in the wild, translation, object recognition, image segmentation, image to text or natural language processing, commercial offerings are leagues ahead of what free software can do.

If we look at one of the most impressive AI demonstrations in history, GPT-3, it is not apparent whether open source can even replicate it because with AI, unlike in the past, time and skill is no longer directly fungible with money. I would argue the concentration of such capabilities to Microsoft and Google servers is a threat to the ideals of free software as great as any it has seen before. Yet, relatively little attention is spent there because people are too focused on yesterday's problems.

This concentration is difficult to avoid because current algorithms require large amounts of data and computing ability, which only large corporations can marshal. Mozilla is far from perfect but despite their many stumbles, they're the only large organization seriously attempting to address this imbalance. As much as these algorithms are marketed as AI to users, ML is better thought of as libraries, in the line of ffmpeg, to programmers. Mozilla still do seem to care about creating a local-first offering. If everyone stops using them then what is gained exactly?

Re: Firefox Voice

#123

Post by nl8ghddbnw »

joecool1029 wrote: post #104
adito wrote: post #102

Earlier quoted context omitted.

That's first time I heard that, not to mean that I didn't believe you, but what's the exact text you sent that is made them censor and banned you?

Screenshot of the text I sent: https://ibb.co/X5LB4qp

Issue 1: "Some fields, such as "URL" and "Email Address", are privacy-sensitive and are only visible to users with minidump access."[1] - So yes, you should not send crash reports when you are dealing with sensitive data.

Issue 2: That is unfortunate and should be fixed.

Issue 3: The issue for Fenix is not 5 years old, it is now one year old, see [2].

Regarding your ban:

- "I will not apologize for being spicy about these things"

- "It is a fact that you've made it intentionally difficult to use your own product to replicate functionality present in most browsers"

-> See the Mozilla Participation Guidelines - "Be respectful in all interactions and communications, especially when debating the merits of different options." [3] I guess you could have phrased things differently. I will not judge if this is "enough" to ban you, but I do also not know what you were posting before this specific comment, you mention additional comments.

Sources:

[1] https://developer.mozilla.org/en-US/docs/Mozilla/Projects/Cr...

[2] https://github.com/mozilla-mobile/fenix/issues/4337

[3] https://www.mozilla.org/en-US/about/governance/policies/part...

Re: Firefox Voice

#124

Post by Aachen »

> When you make a request using Firefox Voice, the browser captures the audio and uses cloud-based services to transcribe and then process the request.

Is it that hard to do local processing, either due to computational power or storage requirements? Or is it just more convenient for them to do it this way?

Edit: this comment in another subthread kind of answered the question: https://news.ycombinator.com/item?id=24098950

If I'm drawing the right conclusion, it's a bit of both: hundreds of megabytes of storage is fine for most people but not everyone, and while I probably wouldn't listen to the latest and greatest artists (and binary diffs are a thing, small additions aren't that large), it is convenient for devs to just push it to a server and be done rather than pushing model updates to everyone all the time.

Edit2: https://news.ycombinator.com/item?id=24096836 Wait, what?! The data is all sent to Google? I was thinking of using this for their sake (opting into using my data for common voice) but this is an instant deal breaker.

Re: Firefox Voice

#125

Post by est31 »

panpanna wrote: post #111
AsyncAwait wrote: post #107

I see a lot of skeptical voices here, (somewhat warranted, given it's a voice assistant technology), but the fact remains that if we want open, on-device voice recognition, we'll have to do the work and donate sample data. This extension is trying to provide some useful functionality in the hopes that Mozilla gets more data for https://commonvoice.mozilla.org I'd at least consider recording your voice, especially if…

> if we want open, on-device voice recognition, we'll have to do the work and donate sample data. Fair enough, but is any stopping Mozilla from _also_ selling your voice data to third parties including advertisers and commercial ML interests? (Asking this because Firefox has started sharing our data with leanplum)

Right now you can only help the STT engine by contributing to commonvoice.mozilla.org, where the samples are published to the world. The add-on will now, if you opt into it, keep the data in only Mozilla's and Google's hands. Mozilla has an agreement that Google won't keep the data, but even if Google doesn't comply by the agreement, the number of parties with access to the data will be lower than "everyone".

Re: Firefox Voice

#126

Post by est31 »

scrollaway wrote: post #108
posguy wrote: post #88

Earlier quoted context omitted.

Mozilla DeepSpeech is trained on about 2000 hours of audio that is mostly spoken by American males. It has little ability to handle noise or accents and has a 5.97% Word Error Rate on LibriVox (which is noiseless, plain spoken english). Meanwhile, Google, Microsoft & IBM have tons of fresh audio coming in constantly to use in augmenting their models. Baidu was able to build a competitive English Speech to Text model…

There's tens of millions of hours of video content out there that has been subtitled pretty well, and I'd wager a lot of it is under usable licenses for Mozilla. Has that been considered?

If you know such sources, file an issue, and better yet, download the video content yourself and publish a dataset.

But note that raw video content is not training data. It has to be segmented to be in short enough parts for training (few seconds), the subtitles have to be aligned to match what's said precisely, and one needs to balance the data, e.g. when 90% of speakers are men and 10% are women, you have a problem.

Re: Firefox Voice

#127

Post by input_sh »

dhaavi wrote: post #80

"We’ve instructed the Google Speech-to-Text engine to NOT save any recordings." Hahaha! :D Thanks for the good laugh.

Bigger players have the leverage to get companies to do something they don't do out-of-the-box. They can contractually oblige them to do that, as well as sue each other if one side breaks its part of the deal.

Re: Firefox Voice

#128

Post by est31 »

Cybiote wrote: post #122
panpanna wrote: post #111

Earlier quoted context omitted.

> if we want open, on-device voice recognition, we'll have to do the work and donate sample data. Fair enough, but is any stopping Mozilla from _also_ selling your voice data to third parties including advertisers and commercial ML interests? (Asking this because Firefox has started sharing our data with leanplum)

It still remains a risk well worth taking. With Mozilla, it is merely uncertain but for just about every other company, it is all but guaranteed. You can plainly see this philosophy in the industry's naming sense, where super-computers of yester-year are relegated to "edge" roles. As of today, the open source and free software equivalents to machine learning and AI products are sorely lacking when compared to commerc…

> it is not apparent whether open source can even replicate it because with AI, unlike in the past, time and skill is no longer directly fungible with money.

Indeed. The creator of LuaJIT is only a single, very skilled, person with a desktop computer. People like Fabrice Bellard can produce gigantic amounts of FLOSS source code. Yes, those are only examples, but people with their skills and motivations to build FLOSS software will need access to lots of money in order to be able to build ML models.

Re: Firefox Voice

#129

Post by dreamcompiler »

AsyncAwait wrote: post #107

I see a lot of skeptical voices here, (somewhat warranted, given it's a voice assistant technology), but the fact remains that if we want open, on-device voice recognition, we'll have to do the work and donate sample data. This extension is trying to provide some useful functionality in the hopes that Mozilla gets more data for https://commonvoice.mozilla.org I'd at least consider recording your voice, especially if…

> if we want open, on-device voice recognition, we'll have to do the work and donate sample data.

We absolutely will not. The only reason people believe this is that they've forgotten how to do speaker-dependent recognition (SDR), which is more accurate and more secure anyway. We were doing SDR in the 80s with 1/1000 the CPU power and 1/1000 the memory.

SDR does require an initial training session, but once that's done any modern computer or smartphone should be able to handle it locally with no cloud server environment.

Re: Firefox Voice

#130

Post by eganist »

dreamcompiler wrote: post #129
AsyncAwait wrote: post #107

I see a lot of skeptical voices here, (somewhat warranted, given it's a voice assistant technology), but the fact remains that if we want open, on-device voice recognition, we'll have to do the work and donate sample data. This extension is trying to provide some useful functionality in the hopes that Mozilla gets more data for https://commonvoice.mozilla.org I'd at least consider recording your voice, especially if…

> if we want open, on-device voice recognition, we'll have to do the work and donate sample data. We absolutely will not. The only reason people believe this is that they've forgotten how to do speaker-dependent recognition (SDR), which is more accurate and more secure anyway. We were doing SDR in the 80s with 1/1000 the CPU power and 1/1000 the memory. SDR does require an initial training session, but once that's do…

Who's "we" in this context? Because just below you, HN has comments from willing donors.

My point being that while there may still be a market for SDR, there's a broader market for speaker-independent recognition (SIR) simply because people want the tech to just work rather than feel like they messed up training the device when the device can't recognize them.

Post reply on HN