Creator of Linux virtual assistant blames 'patent troll' for project's death | Hacker News Reader

Creator of Linux virtual assistant blames 'patent troll' for project's death

theregister.com

218 pointsLorenDB109 comments

Software patents are a huge drag on small companies and innovative large ones, it's just that large companies are better able to handle the costs. Our world would be better if we ended software patents. Check out FSF's campaign to end software patents: https://endsoftwarepatents.org.

This isn't a software patent.

Yes it is:

Voice Tech's patents, according to their abstract, involve "receiving audio data from [a] mobile device at the computer. The audio data is decoded into a command. A software program that the command was provided for is determined. At least one process is executed at the computer in response to the command. Output data is generated at the computer in response to executing at least one process at the computer. The output data is transmitted to the mobile device."

https://www.theregister.com/2020/02/12/mycroft_patent_troll/

This is different. This isn't a patent of an algorithm or similar implementation in software. This is a means of controlling a device by decoded voice commands via another device. It's not a software patent in the way FOSS people mean when they say to ban software patents.

Describing a shovel as an "emplacement tool" does not make it not a shovel. Similarly, dressing up a software patent to disguise its functionality as a software patent does not make it cease to be a software patent.

If software that runs on a prior art general purpose computer can infringe a patent, it's a software patent.

If the difference between a device allegedly infringing and not infringing is whether it's running a certain piece of software, then obviously the software is what's patented.

The abstract isn’t the part that matters.

Take patent US9794348[1] which covers some specific heuristics that can be used to decide which voice command to execute. The only remotely interesting part of the claims (the "audio command interface") is definitely software, the detailed description even says that it could expose an API and other applications could use it to expose voice commands.

[1] see https://patents.google.com/patent/US9794348

I wasn’t taking a position about whether this qualifies as a software patent. Just pointing out that the abstract is near-worthless in making any evaluation of a patent.

Formally it's intended as a hardware patent due to involvement of computer. But since it doesn't invent hardware, it's merely invalid patent. AFAIK USPTO bears no responsibility for granting invalid patents.

"it's intended as a hardware patent due to involvement of computer"

All software patents are hardware patents since the program runs on the computer.

That's the loophole that the EPO has been using to allow software patents to be granted anyway, despite their exclusion in the law.

After the failed attempt to codify their practice in law in 2005, the large companies and the patent industry decided to install their own court, the Unified Patent Court, which might go live next 1st March.

They broke the rule of law as they needed the signature of the UK, with the blessing of the European Commission Thierry Breton, the guy who was ex-CEO of Technicolor, who made a fortune with MP3 patents.

How will they enforce decisions of their court?

It's only an invalid patent if you manage to get review before SCOTUS. Otherwise your road ends at the federal circuit which has yet to meet a patent it doesn't like.

Pretty sure you can go up from federal court? It happened.

How isn't this an obvious invention?

But it took a million dollars to defend that it wasn't one patent, from one person.

Seems like a very very real reason to be horrified by software patents.

Can you explain why it's not what EFF means? To me, the part of software patents that doesn't make sense is that you can patent any possible combination of ideas — in this case the combination of using voice and remotely controlling a computer — and it's considered a valid patent.

In my opinion, this is like patenting putting a blue Lego brick on a red brick because nobody bothered to write down or patent that particular combination before. There's no core novelty here, and obviously Mycroft was not stealing profits from Voice Tech Corp.

Honestly the entire patent system should be thrown out / winded down. Nothing but dead weight, especially when huge countries like China completely ignore them.

We can find other ways to finance "innovation".

I honestly can't think of a single incidence of the patent system promoting innovation in the last 20 years.

For small companies patents are basically worthless, a big company can just play legal ropeadope until they run out of money.

A small company (e.g., a startup) will never sue to enforce its patents because suing is too expensive. Because they won’t sue, they won’t be able to license the patents to anyone.

For a startup, however, patents can still have value in at least the following ways:

1. Patents on improvements may discourage larger competitors from suing the startup for infringing the competitors' patents, because the startup may counter-claim infringement of its improvement patents.

2. A potential acquirer may want to acquire the company for its patents. The acquirer may be an established competitor who does have the resources to assert the patents against other competitors, either offensively or defensively.

3. Patents may encourage investors who see patents as residual assets that would be liquidated if the company fails.

Value (1) illustrates the absurdity of the patent system. Innovation, the thing the patent system ostensibly protects, is put at risk by..patents. And so the value of a patent for inventors is to protect their innovation from...patents.

There are very few places in law where you can legally stake a claim within someone else's territory and exploit it for profit. Our patent system is one of them because it allows one inventor to patent an improvement to another inventor's patented invention. That should (1) encourage early inventors to continue to invest in improving their inventions before someone else does and (2) encourage new inventors to improve on existing, still-patented inventions by giving them ways to profit from their improvements.

To be clear, that anyone can improve upon anyone else’s invention and claim their improvement as their own, to the exclusion of the prior inventor, is what drives innovation.

>I honestly can't think of a single incidence of the patent system promoting innovation in the last 20 years.

Without patents pharmaceutical industry would be nearly impossible, i.e. it's very hard to come up with a compound but very easy to synthesize it as soon as it's known.

I'm not sure that making it harder for people to synthesize medicine is a good thing.

If it costs a billion dollars per medicine to develop someone needs to pay that cost. If you don't have some sort of government medicine research branch with a budget of tens of billions to develop medicines then private companies want a way to recoup their investment and then make a profit.

Is it impossible for government to directly fund research into needed medicines? It’s hard to see a value in this arbitrary middle layer between public interest and research. The amount of money people spend on drugs is staggering.

This is too big of a discussion to have in hacker news comments. But the decline in R&D spending, the amount of grants drug companies accept from the government, the amount of government funded research that goes into new medicines, the current state of healthcare spending in the US compared to other countries, the politics of it, the success of companies such as Cost Plus Drugs, and the amount of new medicines that come from countries that have much higher medical regulations all point to the fact that this is a bad system. I have worked on a few healthcare projects, and the general consensus is that it is unsustainable to have a system where health outcomes decline, even if profits are going up. It has been prevailing wisdom that patents help recoup the (admittedly enormous when you factor in clinical trials) cost of developing new drugs, but it has put the US especially into a really awful healthcare situation compared to other nations.

Long story short, there are enough sick people to provide medicine to and have a profitable business through a high volume of care, and governments are much better at providing long term financial stability than private corporations to burden the necessarily regulatory costs.

The pharmaceutical companies routinely pay companies not to produce generics of drugs that aren't patented, so I'm a bit confused by this statement.

https://www.ftc.gov/news-events/topics/competition-enforceme...

https://www.justice.gov/opa/pr/pharmaceutical-companies-pay-...

https://www.justice.gov/usao-edpa/pr/three-generic-pharmaceu...

Pharmaceutical research can be funded with public money. Bounty programs, grants, et cetera. A lot easier to manage those than the obtuse patent bureaucracy.

Patents aren't there to promote innovation, they usually do the opposite so the original inventor can make bank. There's no competition or incentive to perfect your product because you're a state enforced monopoly.

Take FDM 3D printers for example. Stratasys figured out and patented the first one almost 35 years ago and then proceeded to do fuck all with it, selling mediocre garbage for sky high prices. After it expired in 2009 3D printing rapidly became a whole industry with dirt cheap hardware, enabling millions to make things never before thought possible. We could've reached that point in the early 90s if it weren't for patents.

Sure one might say, it also forces the patentee to disclose the inns and outs of the invention, but frankly those patent descriptions are often as bare bones as possible and you'd get there reverse engineering any product far sooner than 20 years.

I have a family member who used to work in pharmaceutical R&D. I personally know nothing about the industry so maybe it's nonsense, but she once told me that nobody would make anything in that industry if it wasn't for patents. R&D costs are astronomical. Once you release a drug, it needs high margins in order to make back the R&D costs. But the problem is: once you release your drug it immediately gets reverse engineered and other facilities can mass-produce it and undercut you hard.

Anyway it's likely I misunderstood something there since again I know nothing. But after hearing that, I considered that while software patents are obviously bad, maybe they make some sense in other industries.

Pharmaceutical research can be funded with public money (and often already are to some degree). Bounty programs, grants, et cetera. A lot easier to manage those than the obtuse patent bureaucracy.

Why do people comment so confidently on industries they don't understand?

Why do people jump to defend monopolistic, price-gouging pharmaceutical companies?

The parent didn’t defend anyone.

Yeah because the parent comment had nothing material to say. They basically just told me to shut up for no reason other than "you have no idea what you are talking about (but I do)".

What don't you like about what the commenter said? Here's support for his comment:

https://cepr.net/we-can-develop-new-drugs-without-patent-mon...

> While drug companies do operate to make a profit, the part of the story that Nisen misses is that the profit does not have to be gained through patent monopolies. Suppose that the U.S. and other governments put up research funding, which private corporations like Gilead could bid on based on their expertise and track record. In this case, a condition of the research is that all patents would be in the public domain (the companies were already paid for their work) and all results would be fully public as soon as practical.

The above comes from Dean Baker, who seems to be confident and credentialed.

https://cepr.net/staff-member/dean-baker/

https://en.wikipedia.org/wiki/Dean_Baker

So help us understand then why this industry could not exist without patents?

Because the requirements for anything health related are so absurdly high, the investment required to put something on the market is also absurdly high. You can’t make that investment if the next guy can then just go and sell your product for a quarter.

This still makes me wonder why a government could or should not invest in it.

Why do people in the industry claim nothing would be invented without patents?

They don't know a world without such patents and they don't know the people and companies who would do research in such a world.

That's a claim like nobody would be a fire fighter if it wasn't a paid job. Guess how reality is really like.

What about all the underpaid people in healthcare? By that mindset those people wouldn't exits.

Figurative understanding and truth is mutable. Society put people on the moon when agency is put to it.

Why do people comment about the way society works when clearly agency is monopolized to focus on fiat wealth targets, not real work of value.

Easy enough to just federally fund a bounty based off qalys saved or something similar. The acts of innovating and manufacturing don't need to be linked.

Software decompilers are the software industry's reverse engineering problem, so internet services are the solution, like those running from the cloud/data centres.

Of course patents also create monopolies and because no one wants to go backwards, monopolies never get challenged or challenged with little consequence so you can hold many country's and their population hostage quite legally.

Thats and actual physical product produced and sold and bought. Why can't the patent system protect only these people?

>especially when huge countries like China completely ignore them.

Excellent point, and it's the same with some proposed AI Laws.

We need a way to encourage innovation and patents do work for that. The issue is not patent per see, the issue is the US patent office.

There is next to zero proper due diligence done for a patent to be awarded and there is no consequence to the office for issuing invalid patents. Just significantly raising examination fees, getting the patent office on the hook for issuance of invalid patents and fining companies filling for frivolous patents would solve most if not all of the current issues.

I leave investigating why it's not done as an exercise to the reader.

That campaign is an initiative from the FSF, by the way, though it's not surprising the EFF may also be promoting it.

Oops, good point! EFF also has some good articles about software patents: https://www.eff.org/issues/patents.

No problem. I point it out because I think the FSF often deserves more credit than it gets.

The actual patent seems to be https://patents.google.com/patent/US9794348B2/en

It's incredibly generic; it basically just describes the computer from Star Trek: "computer, where is captain Picard?" There's a little bit more to it, but not much. There's also a number of very similar patents in the Patent Citations section with only superficial differences, so it's not even the first patent on this.

Nothing was innovated here beyond the obvious; this patent shouldn't have been approved. The idea is that either 1) investments are protected by granting exclusive rights for a limited period, and 2) the public benefits by making the innovation public so they can benefit after the exclusive rights expire. But neither is the case here: there was no innovation here beyond writing the patent, and the public doesn't benefit because it's a generic high-level description like we've seen in science fiction for many decades.

Whether patents are even a good or bad thing in the first place doesn't even come in to play.

IIRC the patent office cracked down on bullshit like this around a decade ago? This one precedes that though.

https://www.pcmag.com/news/white-house-plan-cracks-down-on-p...

I'm not sure how this idea is even considered unique in the computing world as we have been able to talk to computers for decades. For example, Dargon's Naturally Speaking is several decades old and it lets you talk to the computer with the computer translates those voice commands into computer actions. It's hands free, voice controlled computing.

The patent is specifically about heuristics to pick between various commands. For example if I say "turn the light on" while in the kitchen it would turn the kitchen light on, if I am in the bedroom it would turn the bedroom light on. Also depending on whether I or my wife say "good night" it might turn off different bedside lights. Still pretty obvious to anybody who has read any sci-fi, but it's not about voice commands in general.

I posted in another thread but I'm curious what happens when you throw a large language model into the mix that can "just do this". Suppose you prompt it with "input: <user>, <query>, <mic intensities in locations> output: an appropriate api request in <format>".

ChatGPT and the other GPT variants can already parse natural language requests much better than the incumbent systems can (for now). My Alexa device seems incapable of understanding that I have a "bedroom" area defined in Philips Hue, yet clearly knows individual lamp names.

You need to define the room in Alexa ecosystem then this works fine. You tell it a room, and what devices are in it, then it’s “smart” about it. Don’t have Hue specifically, but Alexa doesn’t import room springs from other ecosystems I’ve used, you have to redefine them.

I would love to see a broad effort to certify, promote, and support open source privacy respecting devices from smart watches like pinetime, open laptops and phones, routers, even CPAP devices.

it's incredibly important for us to have maintainable open alternatives to the proprietary hellscape I see even when trying to purchase the most mundane devices.

some markets lacking open alternatives now:

security cameras

smart speakers

CPAPs

working OSS phones

wireless routers (shipped with open firmware)

automotive diagnostic/programming software

smart TVs

smart scales

air tags

smart thermostats

personal weather stations

Here are some ways you could rank devices:

Open API with documentation?

Open Source Firmware that can be deployed and updated by the user.

Open Source hardware specification (think RISC V)

Does this device produce data in an open format or standard?

Does the device send data to the manufacturer or third parties?

Does this device depend on cloud infrastructure the end user cannot control or replace with their own?

Does this device require you to install a mobile app to configure or manage it?

Can the device be repaired or upgraded?

What is the typical support lifetime by the manufacturer for software updates?

What is the MTBF?

Does the device have a secure supply chain?

Are cad files available to support modding and repair?

Have taken a look at Mozilla's Privacy Not Included project? https://foundation.mozilla.org/en/privacynotincluded/

> CPAPs

> automotive diagnostic/programming software

The regulatory evidence you need to submit, the standards you need to comply with and the quality control you need throughout the whole manufacturing process makes it near impossible to deliver community driven open-source medical devices unless it's a 'wellness' (not medical device) or maybe class I medical device.

You could argue that Manufacturers could design open hardware and let people modify code at will on their custom CPAP but it would be very difficult to navigate legal liability. If someone modifies their custom CPAP and makes a firmware change that drastically increases pressure and a patient suffers barotrauma who is liable?

The same can be said for safety critical firmware in automotive (excluding infotainment etc.). Diagnostics/Logging sure have at it, but actual firmware changes I would prefer that we let manufacturers do that. A 4 tonne vehicle at 100 Km/h can hurt a lot more people than just a tinkering driver.

Over the weekend my wife was saying how nice it would be to have a "hey google" that wasn't sending data to Google all the time. My wife is not a tech person, but she was really creeped out be having a conversation with a daughter about music, and then having the artist my daughter mentioned show up in her YouTube music feed. Could have been coincidence... but...

I never had to solve that problem in detail, but - check the software you have installed, linked in any way to a microphone, and its permissions, and all the "contracts" (licences) you may sign or have signed.

Some who are spared from that phenomenon of eavesdropping have followed that path.

It's more likely that the daughter played music from that artist on YouTube, and your public IP address was associated to it in YouTube's data.

Still sketchy, just less mysterious.

Nah I have the same regularly when I only talk about things with a friend. I think devices are listening more than disclosed. I mean, you can hide it by doing the analysis on-device, and just sending some encrypted tag of a few bytes whenever there's an API message being sent.

But it could just be a coincidence of course

that could also be a similar story- device proximity and contacts being used to infer common interests between two people. If someone's talking about a topic (even if it's a long-standing interest) it's highly likely they've searched for something related to it recently.

> Nah I have the same regularly when I only talk about things with a friend.

Sure, but did your friend google those things at any point (before or after your conversation)? Do you ever connect to the same wifi access points? Are your phones ever in physical proximity?

Almost certainly your answer would be "yes" to most of those questions, which also explains how you get the suggestions about things you "only talked about" with your friend.

I agree that this is not a coincidence, but the explanation is rather simple, really. If there is a user A who googles topic X, and there is another user B who appears to be their "social link" (via device proximity, occasionally utilizing same wifi access points, sending emails to each other, etc), it makes sense to suggest topic X to user B.

Well, once I was making a peanut butter sandwich. Next ad I saw on my phone was for an organic brand of peanut butter. I never said or searched for anything related to that. So my theory is that the phone has an undisclosed odor sensor in it that is sending back data.

My point being, that if you have a large enough data set then you can find all kinds of coincidences within that set. I.e., I'm sure you've heard of the comparisons between Lincoln and Kennedy (i.e., Lincoln was killed in the Ford theater, and Kennedy was killed in a Ford). This is also known as the Texas Sharpshooter's Fallacy.

If your friend searches it, it can get recommendwd to you since they do social graph analysis.

Either devices are listening more than disclosed, or you're listening more than usual. Perhaps, after talking about things with your friend, you become more perceptive to similar things that occur in close temporal proximity.

> regularly when I only talk about things with a friend

I've read the other good responses to what you've said, but nobody said this:

it is very easy to become part of a trend or fad without realizing it. Like you could have an avocado toast at your local cafe and fall in love with avocado toasts without realizing that the cafe is already part of a national avocado toast trend. So you only talked about that avocado toast with one friend on your weekly wilderness hike, yet somehow google suggested you might like avocado toast.

mentioning it because in this case there is no direct link, but it isn't coincidence either. So many things I've fallen in love with and taken up turned out to take off in a trend, but I didn't cause the trend nor notice that's how I heard about it.

It probably saw the daughter accessed stuff from a home IP address so suggested stuff to every device on that same IP.

And now due to this patent trolling the Googles of the world don't have to worry about people side stepping their surveillance and cutting into their profits. Not that there's a conspiracy. But it really sucks that these deeply entrenched mechanisms in our economy just keep taking freedom from surveillance away from people.

It's not coincidence. Google does this with articles in the Android assistant when you do searches on a separate device. Everything that can be linked to a common owner will have ads blasted to it based on their consumer profiling.

N=1, i was talking to my father over Meta’s messenger about my plans for the future. I mentioned doing research in $Country, a visit to the grocery store later and the ads in my messenger app were for a research degree in $country.

We were even speaking in a mixture of english and $language, and it still figured stuff out.

Unfortunately i cant get my friends and family to adopt a more privacy friendly platform.

[Edit: It appears there were at least two suits filed by the same patent troll. The first was in east Texas and was dropped by February 13, 2020 (https://web.archive.org/web/20200305135828/https://mycroft.a...). However, five days later, on February 18, 2020, the same patent troll entity filed a second suit in Kansas City (https://www.bizjournals.com/kansascity/news/2020/03/30/mycro...). According to the latter article, their projected costs to fighting the suit were already at least ~1/4 million$ three years ago (March 2020).]

It's possible the patent troll spooked some investors and caused them to back out.

Unfortunately the same patent troll sued twice. Mycroft's blog posts on the second suit:

https://web.archive.org/web/20200807211354/https://mycroft.a...

https://web.archive.org/web/20200516014621/https://mycroft.a...

Have you read the article text or just the headline? The article says the same thing you're saying (trill dropped the case) & cites legal costs already incurred by then as being a burden on mycroft.

---

[Edit] Ok you've edited so my reply no longer makes sense, but one case or two there would be legal costs to mycroft regardless.

Mea culpa for kneejerk commenting. The article is a little confusing though; what actually seems to have occurred is:

1. Patent troll entity filed a suit in east Texas, c. February 2020 (https://web.archive.org/web/20200410225148/https://mycroft.a...).

2. Same patent troll entity dropped the first suit, also c. February 2020 (https://web.archive.org/web/20200305135828/https://mycroft.a...).

3. Same patent troll entity filed a second suit in Kansas City on February 18, 2020 (https://www.bizjournals.com/kansascity/news/2020/03/30/mycro...) (https://web.archive.org/web/20200807211354/https://mycroft.a...) (https://web.archive.org/web/20200516014621/https://mycroft.a...).

This second suit seems to have substantially cost Mycroft.

According to a subsequent suit filed by the patentee's attorney, the patentee voluntarily withdrew the suit in the Eastern District of Texas after its lawyer and the lawyer's family were subjected to a targeted campaign of harassment and cyber attacks. [0] The patentee then re-filed the case in Mycroft's home state of Missouri. [1]

[0] Tumey L.L.P. et al v. Mycroft AI Inc. et al Case No. 4:21-cv-00113 (W.D. Mo.) Complaint https://fingfx.thomsonreuters.com/gfx/legaldocs/egpbkqlzqvq/...

[1] Id. at 13.

> Voice Tech Corp., based in Tyler, Texas, filed suit Feb. 18 in federal court in Kansas City, accusing Mycroft of infringing on its patent for using voice commands from a mobile device to remotely access and control a computer.

I'm sure there's some prior art to be found here, but the lawyer fees are still going to be a lot to fight it. Sounds like that's what doomed them.

Tyler Texas Patent Trolls: https://www.youtube.com/watch?v=d8XknFl1l_8

Sounds like it would be really, really easy to filter out these trolls if Texas courts weren't simply corrupt. I wonder how the payment for these judges is arranged.

They say the judge is the father-in-law

I was excited about this project, so I'm saddened by this news.

Having said that, I'm a little confused by the outcome here. What I really want is a pre-built piece of hardware that runs the Mycroft open source project, so I don't have to go figure out which speaker and microphone array works with whatever Raspberry Pi clone I can actually find in stock at the moment. I would be a little disappointed if there were no longer an entity working on the software itself, but honestly all I ever use this type of device for is 1) setting timers 2) checking the weather 3) playing music, so assuming those things work reasonably okay with the current software, I'm not really in need of further innovation here.

Back to the hardware - what am I missing? Can I buy this? The Mark I seems to no longer be manufactured and the Mark II is what seems to have now become vaporware because of this litigation. I understand it might not be a loss leader like a Google Nest Home Assistant Hub Device (I've given up on actually understanding Google's naming anymore), but it really shouldn't be that expensive for a "dumb" smart speaker. Does anyone make this? I'd prefer no proprietary hardware, but I'd accept a somewhat-closed piece of hardware if it could run software of my choosing.

According to the page, you can still buy the Hardware, it's made by a "partner" and it's self funding

"Which brings us to the present day. We will still be shipping all orders that are made through the Mycroft website, because these sales directly cover the costs of producing and shipping the products. However we do not have the funds to continue fulfilling rewards from this crowdfunding campaign, or to even continue meaningful operations."[1]

1.https://www.kickstarter.com/projects/aiforeveryone/mycroft-m...

You're right - I read the Register article linked here and jumped straight to the project homepage without realizing the stock listed there was a separate source than the Kickstarter described in the article.

Had the MyCroft team made a dedicated funding project for their legal defense, I would have gladly funded it as would many others, because fuck patent trolls.

Too many open source projects deal with this sort of thing in secret until it is too late.

We really need an open source project patent troll defense fund that reigns millions in legal hell down on the first to try to hit any company protected under it including counter-suing for damages until it sinks them. Also hire private investigators to find the real names and backgrounds of every asshole on the other side of the suit, and anyone who supports them. Make everything public to the furthest extent of the law and socially bankrupt them all.

We need public scorched earth to send a message anyone else that tries to hit any other protected projects.

No direct mentions of the patents in the article, but indirectly through an older article[0]. The patents in question seem to be [1][2] (both the same?!).

> A method of using voice commands from a mobile device to remotely access and control a computer. The method includes receiving audio data from the mobile device at the computer. The audio data is decoded into a command. A software program that the command was provided for is determined. At least one process is executed at the computer in response to the command. Output data is generated at the computer in response to executing at least one process at the computer. The output data is transmitted to the mobile device.

How can a patent like this pass? This is absolutely ridiculous. Patents relating to CS are utterly broken, people need to demand the government to step in since it is impossible for small players to fight them, hence it stifles innovation.

0. https://www.theregister.com/2020/02/12/mycroft_patent_troll/

1. https://patents.google.com/patent/US9794348

2. https://patents.google.com/patent/US10491679

> But what truly killed the company and product, he claimed, were expenses related to ongoing litigation.

Some of this cost includes defending against the lawsuit alleging that Mycroft and its principals initiated or called for a campaign of cyber attacks and harassment against the lawyer for the patentee and his family. [0]

[0] See, e.g., Tumey L.L.P. et al v. Mycroft AI Inc. et al Case No. 4:21-cv-00113 (W.D. Mo.) https://fingfx.thomsonreuters.com/gfx/legaldocs/egpbkqlzqvq/...

I'm super bummed to hear this. I got my order in very early and am fortunate to have actually received mine. The unit itself is pretty bulky, it turns on/off lights via home assistant well, but feels pretty half baked because it doesn't do much more than that.

Very disappointing to see. I knew something was up since Mike Hansen, the engineer maintaining their Mimic3 offline TTS model, mentioned being let go in the open GitHub issues.

https://github.com/MycroftAI/mimic3

Mike seems to have kept the effort going with his larynx2 project. I was thrilled at the results their model produced so I hope to see more from it.

https://github.com/rhasspy/larynx2

I liked how the GNOME community dealt with their patent troll. IIRC the patent is no longer allowed to be enforced.

I do not understand why he does not move the project to, say, France and show the middle finger to the patent troll.

Sounds like the litigation has been dropped, but the $1m in attorneys fees already paid was a loss too large to bear.

If you want to sell to people in the US, then US patents still matter.

This is a shame.

Doesn't US businesses have legal insure to cover situations like this?

there isn't a lot of "litigation fee insurance" that is affordable.

How would that work. This is a kickstarter project

Why not come to the EU? That simple.

That's very sad

Ah well, sad to hear. I've tried it a while back and the architecture with skills being separate diy-able scripts looked promising, but overall it wasn't all that usable.

Eventually it would've been steamrolled by a GPT based assistant anyway.

I’m trying it right now and the initial results seem promising. The openai whisper voice to text works nearly flawlessly and everything else seems to be falling into place. I’m thinking of using local gpt or similar models to classify and then distribute requests to more programmed scripts.

Are you just padding input windows to 30 seconds, or have you come up with a smarter way to handle "streaming"?

Just padding, there's a built-in function which just does it. That is definitely in the realm of "problems for much later", the bits that interest me are what to do with the input, and 30 seconds to talk is more than enough to accomplish what I'm thinking about.

A physical button for input is what I'm thinking as a side project, I like that much more than always-on listening. There are also lots of much less heavy weight voice to text which could do streaming listening waiting for a key phrase which when found could start recording and send the output up to the better model. I don't know what they are, but I know they exist.

Reply on news.ycombinator.com