Live data from Hacker News

Mozilla's Speech-to-Text Engine Is at Risk Following Layoffs

phoronix.com

21–30 of 79 posts

Re: Mozilla's Speech-to-Text Engine Is at Risk Following Layoffs

#21
post #11

I’m glad to see the healthy skepticism towards Mozilla lately, they have long been a sacred cow in tech. Just because a company’s mission is positive or aligns with ones values does not mean the company itself is good or is actually carrying that mission out.

Mozilla senior leadership have also been caught up in the politically correct mantra that pervades much of silicon valley, which imho is a huge distraction from making great products. Case in point: https://mobile.twitter.com/newstephen/status/106048929137627...

Re: Mozilla's Speech-to-Text Engine Is at Risk Following Layoffs

#23
My quick test of DeepSpeech, as posted on Facebook back in April:

--

I just ran a speech-to-text converter on a very clear clip of former Doctor Who actor Tom Baker talking in an interview.

The DeepSpeech converter uses the very latest AI deep-learning advancements to 'listen' to the audio and output the spoken words as text.

After 3 long minutes of running it on a 30-second clip, it printed out its interpretation:

"hooloomooloo how booboorowie i have a honeymoon"

Re: Mozilla's Speech-to-Text Engine Is at Risk Following Layoffs

#24
post #3

Semi-related, but does anyone know if Firefox Send is on the chopping block? They took it down before the layoffs due to security concerns, and now I'm worried it won't be coming back online. Every alternative I've been able to find has been either poor quality or too shady for me to ever actually want to use.

I don't know if Mozilla plans to bring it back but I know the solution is just around the corner: host it yourself. If you only want to do things like this you could use any SBC - Raspberry Pi etc - running some distribution of some operating system you prefer - Linux, *BSD, etc - hanging off your own internet connection. Either run something cloudy - Nextcloud, Owncloud, Syncthing, Seafile - or find (or make) a prog…

>Why pay for Dropbox when you can use your very own box with far more storage for far less money?

Managing backups and data integrity is no joke. Almost everyone does a shit job and don't know it until the data is gone.

Re: Mozilla's Speech-to-Text Engine Is at Risk Following Layoffs

#26

Earlier quoted context omitted.

Agree, but you seem to be turning downvoted-transparent. I am also always surprised that Mozilla receives so much praise for its technology stack, when their core source code for Firefox and Thunderbird is such a buggy pain to work with. I understand that they inherited a legacy codebase, but I'm not sure that playing catch-up with Google on (in my opinion) useless features like WebUSB was good technical management.…

Maybe I'm unlucky, but any company I worked in has that legacy codebase which kind-of works, has bugs, and that nobody wants to refactor - and I don't blame them, refactoring code which had years of debugging is not always a good idea if you don't have unlimited money.

I also wouldn't blame them for not fixing old bugs if they had a shortage of developers. But letting critical security bugs go un-patched, so that your team can work on obscure shiny new stuff instead, that looks like wrong priorities to me.

Nobody would switch to Firefox for WebUSB, which is why I chose it as the example. There's only crickets on Twitter: https://twitter.com/hashtag/webusb

Instead, people switch to Firefox for privacy, safety and reliability. Mozilla should prioritize those areas highly. But in the past, they haven't. Instead, they burned a lot of money on acquiring 3rd party services that many Firefox users see as nothing more than unnecessary bloat. Like $30 mio for Pocket.

That's why I point the blame at their management, not their code base or developers.

And just FYI, I'm pretty sure I could hire a world-class team and build an excellent Speech-to-Text Engine just from Mozilla's management budget for a single year. Because AFAIK, their (in my opinion incapable) management pocketed $80 mio in 2018.

https://assets.mozilla.net/annualreport/2018/mozilla-fdn-201...

Re: Mozilla's Speech-to-Text Engine Is at Risk Following Layoffs

#27
post #22

Am I the only one trying to figure out why Mozilla even has a "speech-to-text" team?

You've actually touched on something very important here: Mozilla seems to be getting its fingers into more than it can handle. It wants to do everything but ends up doing a lot of things fairly poorly.

I switched away from Firefox to Brave yesterday. Became tired of how Firefox threads would consume 99% of my CPU endlessly until they're manually killed. This happened about four or five times a day, I'd only notice when performance of everything else tanks. I wish Mozilla would direct more resources towards Firefox than Lockwise and all the other half-baked junk they're spending developer time on.

Re: Mozilla's Speech-to-Text Engine Is at Risk Following Layoffs

#28

My quick test of DeepSpeech, as posted on Facebook back in April: -- I just ran a speech-to-text converter on a very clear clip of former Doctor Who actor Tom Baker talking in an interview. The DeepSpeech converter uses the very latest AI deep-learning advancements to 'listen' to the audio and output the spoken words as text. After 3 long minutes of running it on a 30-second clip, it printed out its interpretation: "…

When I tried it out with English and German phone recordings, it was working competitively. I would have ranked it better than Amazon but worse than Google.

Did you maybe not convert your WAV to the correct sampling rate?

Re: Mozilla's Speech-to-Text Engine Is at Risk Following Layoffs

#29

Don't put a lawyer in charge of a tech company.

"But I think most would also agree that if all corporate lawyers, bank lobbyists, or marketing gurus were to similarly vanish in a puff of smoke, the world would be at least a little bit more bearable." I remember just yesterday there was a great article on HN that discussed bullshit jobs and counted corporate lawyers towards the "goons". The rationale was that if nobody has them, nobody else needs them. https://www.…

It's kind of like the military in that sense. Anything 'PvP' falls in that category.

Re: Mozilla's Speech-to-Text Engine Is at Risk Following Layoffs

#30

My quick test of DeepSpeech, as posted on Facebook back in April: -- I just ran a speech-to-text converter on a very clear clip of former Doctor Who actor Tom Baker talking in an interview. The DeepSpeech converter uses the very latest AI deep-learning advancements to 'listen' to the audio and output the spoken words as text. After 3 long minutes of running it on a 30-second clip, it printed out its interpretation: "…

I didn’t know Tom Baker was Welsh.
Post reply on HN