Microsoft buys Nuance for nearly $20B | Hacker News Reader

Microsoft buys Nuance for nearly $20B

axios.com

658 pointsmoritzplassnig397 comments

For those confused, this acquisition is due in large part by Nuance's dominance in healthcare related products. From Nuance's last earnings release[1]:

“We are very pleased with the strong start to the fiscal year, as we delivered revenue and EPS above our guidance range expectations,” said Mark Benjamin, Chief Executive Officer at Nuance. “We continued to advance our strategic initiatives, accelerating our cloud transition across our core platforms in Healthcare and focusing on our AI-first approach in Enterprise. In Healthcare, we saw solid performance in our cloud-based offerings, growing cloud revenue 28% year-over-year. In particular, we benefited from strong performance in Dragon Medical & DAX Cloud revenue, which grew 22% year-over-year driven by the ongoing transition of our installed base to Dragon Medical One, as well as traction in international, ambulatory and community hospital markets. Enterprise delivered another record revenue quarter, up slightly from its previous record in Q1'20, driven by particularly strong demand for our Security & Biometrics solutions."

Nuance has deep relationships built with nearly every health system in the US and beyond. This fits quite well with Microsoft's corporate focus. Yes, Nuance also has a lot of IP, but I wouldn't expect any consumer facing changes (e.g. Cortana) in the near term.

[1] https://investors.nuance.com/download/EX%2099.1%20Press%20Re...

I work in healthcare software and recently did a spike building a voice assistant for the EMR. We compared Google, AWS, Azure and Nuance's voice and intent recognition and Nuance blew all the others out of the water. When it comes to understanding medical terminology Nuance is way ahead of anything other providers have.

Dragon has been around for 23 years and has been THE product for VR in the medical field for at least the last 10 years (from my experience).

Dragon's actually been around since 1982 (39 years). Sad story about the founders, Janet and Jim Baker, though.

https://www.nytimes.com/2012/07/15/business/goldman-sachs-an...

The Bakers sued Goldman Sachs for negligence, intentional misrepresentation and breach of fiduciary duty, which in January 2013 led to a 23-day trial in Boston. The jury cleared Goldman Sachs of all charges.

https://www.nytimes.com/2013/01/24/business/goldman-sachs-cl...

Very sad story. Two pioneers of voice tech had literally nothing for their hard work.

I think in a lot of cases that is because they tried. You can tell 90% of Google's focus is ad consumers when they develop new services, whereas Nuance has sold medical-focused dictation tools for over a decade.

Wow, did not know EMR voice assistant was a thing. Can’t wait till our team works on something like this

Every medical practitioner has been begging for voice EMR for like 10 years. EHR systems are a nightmare to update; a doctor basically needs an intern with an iPad to wander behind them and translate their dictation into EHR. Whoever controls this space will have a close relationship with every doctor and nurse (and any EMRs that don't will seriously falter)

I worked years ago at a major cell phone mfg company, and we were comparing vendors (circa 2013-2014) for the voice recognition/transcription software. I remember two of the leading options were SoundHound (dba then as "Hound", which has since gone to market as a voice solution[0]) and Nuance which was a company not many of us had ever heard of.

At the time, Hound actually was very very good at language recognition and impressed everyone quite a bit. Compared to Nuance, the experience of conversing with Hound was better as I remember. However, Nuance had the edge in language support... while Hound was great for Western dialects of English, and some others, Nuance supported Mandarin. End of the day it was no contest which product we had to go with.

I'm not surprised that Nuance had continued to be an industry leader all these years.

...

0. https://www.soundhound.com/hound

I'm pretty sure Nuance also does the Comcast Xfinity voice remotes. They are actually really well done. And it's a humongous customer to have your claws in.

I have this and it really is quite remarkable. The entire voice integration in the x1 set top box is really polished. Other than the occasional "wow, this is pretty good for comcast!" I hadn't given it much thought until now.

Can you remove the ads from the EPG?

It’s literally better than Siri. I don’t know how, but it is. I’d say it’s even better than Alexa in terms of its ability to recognize things I say.

I think they've been at it for a pretty long time with old products like Dragon. A friend used to work there some years back and said they pretty much perfected speech detection up to some very reasonable error rate. I imagine they've just continued to cover all the dark corner cases and irregularities and accents etc.

The biggest leg up they have is that their product works in a very controlled domain. Siri is out there trying to be a complete interactive AI human and failing miserably. The Xfinity remote just controls your TV and does it smashingly. I found the Alexa-driven voice control on my FireTV to also be a pleasant experience compared to regular Alexa.

Ironically, Siri (as a virtual assistant) came from SRI's Darpa program "CALO" back in the 2000's. SRI also has a speech recognition platform called EduSpeak, and that is the codebase that was licensed to Nuance for their product.

However, after Siri was sold to Apple, I don't think that they retained the EduSpeak portion for speech-to-text. (I honestly don't know - but it seems to me it did not; I don't think Apple wanted to pay that license fee to SRI).

The TV remote only has to understand TV-related things, like inputs, channels, volume, and program names. Siri and Alexa have to understand everything.

I use Siri every day. Former radio guy. Siri is horrible even for me

They do. The amount of magic involved to make it work is amazing from an engineering perspective.

Microsoft already has relationships with Comcast [0] and nearly every major telecommunication around the world.

[0] https://cloudsolutions.comcast.com/apps/64168/office-365#!ov...

And 100% agreed, the voice is great on the Xfinity remote. Was very impressed.

Universal remote control makes all Comcast remotes.

Forum crowds will look to the tech but this is bang on, CEOs are ultimately coin operated and any company would salivate to get their hands on a vendor that is so deeply burrowed into the operations of the healthcare industry.

Yes, that makes sense

Oh wow. The Dragon headset is a very strange product space. It is used by non-tech savvy doctors to avoid having to type into the EMR.

It’s not just the non tech savvy. It’s substantially faster to dictate text than type it, especially if your hands are occupied with the computer doing something else (ie interpreting radiology exams, dictating a treatment course or visit note while simultaneously reviewing labs).

Only if no one reads the output.

I think a lot of big acquisitions make "sense," given the current market.

The most successful companies have lots of cash, high share prices, and amazing cash cows. They could borrow for (almost) free, so resources are practically unlimited. Their R&D is already well funded. Most of their big, growth oriented endeavours are not cash-constrained. There are usually no factories to build or production to scale up.

Google tried "20% time." They tried "let many flowers grow." Those things seemed ambitious at 2007-scale. In 2021 terms... new flowers need to be S&P 500 companies to represent growth, instead of just clutter. "Meaningful growth," for Alphabet, is a big number.

How else does a MSFT, Google or (especially) FB put $20bn to work? Acquiring "just works."

Of course, there are in-house alternatives. Waymo is an in-house investment by Alphabet that's bigger than this Nuance acquisition... especially if you consider the $bns Waymo will continue to need until some unknown future date. Self driving is looking more hopeful (certainly to investors) than it was when waymo started.... but waymo is still a dubious investment.

Consider that Google could have bought any car company, for about as much as waymo will cost eventually. Car companies have loans, so you could quibble the math... but details.

Acquiring is easy. The path of least resistance wins >50% of the time. We have that dynamic here, both in the human/managers sense and in the arbitrage-like incentives in the market currently.

Acquiring is a simple way to show the Board/Shareholders that you're "doing something" - but I suspect it's rarely very successful in the long run. Unless you acquire a business that ACTUALLY provides some synergies you're just on the path of transitioning from a successful company to a poor imitation of Berkshire Hathaway.

It does have the advantage that you can "spin off" your acquisitions once they fail to do anything interesting (though this is more commonly seen in sunset industries/dying companies (see AOL, Compaq, etc)).

I was actually thinking "Satya, you know that Berkshire is you end game here... right?"

I agree, but I think MSFT (and friends) are in this predicament no matter what. What's the alternative?

That said, I don't really think BRK is the end game.

For one thing, Berkshire is kind of an exception. There are plenty of smaller conglomerates that are actually like Berkshire, but most pretend not to be conglomerates. They pretend to be far more cohesive & synergetic than Berkshire.

Also, Alphabet shows that synergies can be easy to find. Youtube, Android... This generation's acquisitions need to be 10X bigger than that. But... these companies are in uncharted waters. No company has wielded free resources at the scale that MSFT now operates. They aren't the only one currently, but they don't have predecessors... unless we go back to VOC or somesuch.

OTOH... Satya is accidentally in the same position Buffet intentionally sought: Sitting on a pile of capital that must be allocated.

> I agree, but I think MSFT (and friends) are in this predicament no matter what. What's the alternative?

Returning money to shareholders is the common solution for when you don't know how to grow more and profitability in the businesses you're already in and don't have any particular advantage in entering new ones. I'm not saying that's what they should be doing but sitting on a pile of capital that you don't know what to do with isn't a new problem that we need to invent new solutions for.

My own uneducated position is that is a new problem because shareholders don't want their money back. Shareholders have been piling money into stocks [1], so if you gave them the money back they would likely just put it right back into Microsoft (or, more likely, it would signal that Microsoft doesn't know what to do with the money, so people would pull out of Microsoft to invest somewhere else).

[1] https://finance.yahoo.com/news/more-money-poured-stocks-past...

"Returning money to shareholders" is largely a consequence macroeconomic policy or state of affairs. It is, evidently, not actually a choice that CEOs currently have in practice. Dividends and/or buybacks are one of the powers shareholders do tend to wield, in practice. They want to invest more, not less.

Also, IDK if it is a common solution. Nothing is really common at MSFT-scale. A free cash flow like Google, Alphabet, etc. is almost unprecedented.

Meanwhile, I do actually think that this is better for shareholders. IMO "Synergies" is a term somewhere between euphemism and a boomerism but for the purpose of "shareholder value" it doesn't matter. At Monopoly/Unicorn/FAANG scale, there are big opportunities for synergy. Think Google-Android.

Why is Nuance being owned by Alphabet less efficient than being traded independently or owned by private investors? Why is Alphabet owning vanguard more efficient than owning Nuance?

The answer to those question can have no actual impact on reality. If the acquired business is cash generative, they can left to their devices. If the parent company doesn't borrow, then "efficiency" never becomes explicit. Explicit efficiency is relative to cost of borrowing. Implicit efficiency is implied by share prices... and at this point things get foggy.

> I agree, but I think MSFT (and friends) are in this predicament no matter what. What's the alternative?

Well, they can just stop growing, and continue doing what works. If that thing stops working (or they have a good belief that it will stop working before too long), they have two options: 1) do nothing, and gradually wind the company down and return capital to shareholders so they can reinvest it elsewhere, or 2) pivot, and accept that the things they are pivoting to will be a rounding error in their finances for years while they grow.

Obviously this is disastrous in our current economic system; a company that tried this would watch its stock price fall into the toilet before too long. But absent that, why not?

Consolidation is what will bring us to a corporate-run dystopia. I would much rather the world be filled mostly with small and medium sized businesses, with every market open to a lot of competition and even cooperation (on standards, not on prices). But I know, that's just a pipe dream, and humans generally suck at cooperation when money is involved.

>> a company that tried this would watch its stock price fall into the toilet.. But absent that, why not?

lol

>> Consolidation is what will bring us to a corporate-run dystopia. I would much rather the world be filled mostly with small and medium sized businesses...

The alternative to that is trust busting, perhaps. I was commenting on the market logic, so to speak. If we're optimistic, maybe it'll be a corporate-run utopia. Zuck's not great, but I think this generation is kinder than the Carnegie/Rockefeller days.

Look... if Bezos, Zuck, and such continue on trend, they'll soon be very rich. Bigger than the Rockefeller. Their companies will be one par with the VOC/EIC in terms of market cap, but I don't know if it's really comparable to that.

Google/FB are sketchy, if trust-busting comes into play. Advertising is sensitive to both regulation and trustbusting. A ban on snooping, manipulation and overly vigorous advertising would hurt advertising. Trust-busting, like separating adwords from google, hurts advertising monopolies too.

Meanwhile, what happens if a regulator messes up and breaks FB? Would the world lack for social messaging media? If Ford stops making cars, fewer cars are made in the world. If fewer FB likes happen, more sploosh sploshes happen and all is well in the world...

More likely though, no help is coming. That being the case, I think the tech bros aren't the worst candidates for trillionaire status. Someone had to be it. I'm glad it isn't the real estate bros.

Well; having been employed at 4 different companies in the past, where those companies were purchased and then shut down by a competitor - I'm pretty sure that consolidation is almost always a bad thing. For the workers, and the consumers.

Apple has had some decent ones. P.A. Semi, notably. And for $278 million dollars, the ROI on that was ludicrous.

But overall, yes. I can think of so many large-scale acquisitions that didn't go so swimmingly.

  - AOL/Time-Warner
  - Ford/its stable of luxury brands like Jaguar and Land Rover
  - Compaq/HP
  - Daimler/Chrysler

It seems like these “big” mergers tend to now show the synergies people promise. Maybe it’s just too much culture to integrate.

Tech M&A guy here.

This is survivorship bias at its finest.

How about Facebook's acquisition of Instagram? Google's acquisition of YouTube? Android? DoubleClick? Amazon's acquisition of Twitch? I could go on.

For the record, yes there are VERY many acquisitions that go wrong, especially when you get to the $B+ value. The parent's characterization seems in line with the "no one ever got fired for using IBM" and that sentiment is grossly unjustified for M&A. There are a bunch of other factors that go into corporate strategy. One example - buying a competitor to eliminate competition and thus protecting future dollars.

> Facebook's acquisition of Instagram? Google's acquisition of YouTube? Android? DoubleClick? Amazon's acquisition of Twitch? I could go on.

These examples are not comparable to what the OP was saying, e.g AOL/Time-Warner, Compaq/HP are mergers of giants with lot of employees.

Skype’s eventual acquisition by Microsoft is an interesting one.

Can anyone contextualize the horse trading that led to that?

>September 2005, eBay acquired Skype for $2.6 billion.

In September 2009, Silver Lake, Andreessen Horowitz, and the Canada Pension Plan Investment Board announced the acquisition of 65% of Skype for $1.9 billion from eBay.

Microsoft bought Skype in May 2011 for $8.5 billion.

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

>Apple has had some decent ones. P.A. Semi, notably

And, let's not forget, NeXT.

Probably one of the best U$400 something million ever spent.

It's funny how there seems to be so little correlation between acquisition prices and eventual value. The PA Semi acquisition was an enormous success at what was at the time a tiny price. NeXT was an enormous success, though $400M at the time was a bet-the-farm price for Apple.

In contrast, while I'm sure Beats has easily paid for itself, $3B was not exactly cheap, and the results were not 10x PA Semi.

I thought about NeXT, but that was a bit of a weird one, because NeXT executives and technology replaced a lot of Apple’s executives and technology. That ended up being a stealth takeover by NeXT.

Based on my personal experience, since 2008 more and more companies are about showing somebody that you are "doing something" with their money. It could be the VCs, it could be the EU grants, it could be the stockholders, but that's the business model you get with the abundance of capital and low interest rates.

I think this was once true, but is significantly less so in the context of cloud platforms. It’s very easy from a business model standpoint for Microsoft to integrate Nuance NLP modules with Azure and start selling access to them at list prices very quickly.

You don’t need to hunt new customers with a marketing plan; you likely already have customers with these needs in your pipeline so it’s a matter of making sure your AEs know what’s happening. Everything is simpler at scale in a cloud business model, which is why these 3 companies in particular are eating the world.

Can't you just acquire a successful business and then do nothing? Just let the business you acquired keep being successful? It's not that different from investing in the stock market, except that you can have more control over the business you acquired if you need to down the line. You also then own their IP, which might be the most important part.

Here's the challenge:

Let's say there is a business that is successful and generating $10B in revenue and $1B in profit and valued at $15B market cap.

Now you want to buy it.

Historically, you will have to pay something like a 40% premium to its market value in order to acquire it. So you will pay $21B ($15B * 1.4) to own this company.

If you "just do nothing," then the company will presumably still generate $10B in revenue and $1B in profit (and really still be "worth" $15B).

So, you paid $21B for something worth $15B, making the shareholders of the selling company very happy and the shareholders of your company sad.

As a CEO, this is a good way to lose your job.

Effectively, you are forced to present (and attempt to execute) a plan for how the combined business either generates more revenue or has lower costs than the two companies did separately in order to get your board's approval on behalf of your shareholders.

I think that’s what GP meant by poor imitation of BH.

Its rarely the syngeries that make it work, instead its often they bought a business they leave alone that has much more room to grow.

IDK if it's rare that synergies work, so much that it's common for synergies to fail.

Consider Google. They acquired Youtube, Android... Google's skillset was perfect for taking these proving concepts and making them 1080px, so to speak. Now, Youtube and android feed users & data to the adwords cash machine. Youtube and android defend the adwords castle, denying competitors. Fantastic synergy.

OTOH, no company will ever find a synergy with ebay. They have spiky bits where companies are supposed to have copulation bits.

Google acquiring YouTube also had good synergy because it caused bandwidth costs to plummet, which YouTube was having a hard time managing as a startup. The company was bleeding money on bandwidth, because they were an end user who had to pay an upstream ISP for transport [1]. Once Google acquired them, suddenly they're on effectively a backbone network and have settlement-free peering with all kinds of other networks. That drove bandwidth costs down to near zero according to one analysis [2].

[1] https://arstechnica.com/gadgets/2015/04/cheaper-bandwidth-or...

[2] https://www.wired.com/2009/10/youtube-bandwidth/

> you're just on the path of transitioning from a successful company to a poor imitation of Berkshire Hathaway.

You mean Berkshire Hathaway, that's current market cap is ~$615B and hails arguably one of the most successful investors of all time as its CEO? That Berkshire?

I don't know about you, but I'd happily be just 1/100th as successful as how that model turned out.

google maps and youtube and instagram were acquisitions

2/3 of those arguably are among the most important parts of their respective companies

Just curious the difference of Waymo and Boston Dynamics.

Is it just me or have speech recognition platforms like Siri actually gone backwards in the last few years (roughly coinciding with the AI/ML craze)?

It feels (anecdata) that Siri is doing a bit better on voice to tex when sending messages but much worse on simple commands like “turn off living room lights”.

This never seems to be a popular opinion on hn, but this is the ultimate Rube Goldberg machine: A complex ai built by a billion dollar company, an enormous amount of man hours, only so that switching a light switch because slower, and less deterministic. I'm certainly not suggesting that there are no valid uses for ML/AI, but digital assistants seem to be an enormous waste.

I'm a fan of Google Assistant's usefulness and I generally agree with this. The Assistant craze seems more justified when you consider that companies like Google and Amazon are using it as a way to build datasets. It's a strategic reserve of data to build other things from (i.e. Google Duplex) and it doesn't matter at their scale if there isn't a vision yet, if a competitor is stockpiling transcribable voice data, they need to keep up.

> The Assistant craze seems more justified when you consider that companies like Google and Amazon are using it as a way to build datasets.

Basically every new goddamn trend that is allowed to occur is about how datasets can be compiled from such marketed options.

That's why we don't have first-class convergence devices yet. It'd likely blend our home and mobile usage profiles too much.

Ignoring the fact that the research isn't only used one for one purpose, a proper assistant is hugely valuable to me.

Simple things like:

"Remind me to X tomorrow at 9am" "Add Y to the shopping list" "Remind me to do Z when I get home"

This is the only way I do reminders now and it's great. It's especially useful while driving, or when I'm in bed nearly asleep and remember something, I can just tell Siri without having to get up and use my phone.

And as far as smart home controls, being able to say "Siri, turn all the lights on" or "Siri, turn the heater on" without having to stop what I'm doing and walk around the house flicking switches is really nice.

Assistants provide a very noticeable QoL improvement for me in many aspects. I think that's more than most other products on the market can say. And that's not even touching on the lives saved from not having to use your phone while driving.

Yes but... Siri is (to me) not there yet. Promising but not there.

Siri is unaware of so many things.

A simple command that would actually be useful (but fails totally):

« Siri, remind me to buy milk the next time I go get groceries. »

And even then... I should not have to mention the part about « groceries ». The request should be perfectly understood with a full stop after the word « milk » and a reminder should pop up automatically whenever I am in a grocery (any grocery).

An even better Siri would also be able to a categorize things properly. For example, if later in the day I say « Siri, I will need chicken, butter, salt and cardamom for my next recipe », Siri should automatically add those to the « milk » next time I go to the grocery.

heh. I'll just say that Alexa rarely misses an opportunity to "remind" me when I need to buy something.

I'm still not very impressed by this, as it can be totally accomplished by a pad of paper and a pen.

I'm believe the most valid use of AI/ML is to perform tasks that people either cannot currently do, or cannot easily do. For example, ai-based up-scaling of old video game pre-renders. It's not really feasible for a person to do this well, unless you simply rebuild everything with a team of artists. And you could argue that up-scaling images for a video game is trivial, since all video games are trivial. But, the point is that the task at hand requires the help of a computer, whereas a to-do list, or using a light switch does not.

Let's see the situations when you are setting those reminders -

1. In bed, all ready to fall asleep, you get up to get your notebook from your desk to add the reminder to your notebook.

2. Cooking with hands all messy, you wash your hands, dry them, go to your desk to add the reminder to your notebook.

3. Driving at 40mph, so you pull over in a parking lot, take out your notebook from your bag and add the reminder.

Yeah, I will take telling Google assistant to do this instead.

> tasks that people either cannot currently do, or cannot easily do.

A todo entry or flipping lights via assistants also qualify as such tasks, if we broaden the definition of "easy". The flow from having the thought of an idea or a song to making a note or playing the song by just speaking out loud is just so convenient, without having to context switch from whatever one is doing. Controlling a set of IoT devices with custom commands is another good usecase.

Of course, not everyone has a workflow where a digital assistant fits well today. However, I expect that their usefulness will increase exponentially with time. We're surely heading to the sci-fi future where each house will have a personalized digital guardian responding to the wishes of the family, Jarvis style, no?

Speech recognition goes way beyond digital assistants. Plain old transcription is super useful, especially in terms of accessibility. I was voice coding for a while because of RSI and even as immature as the tech is, it saved my ass.

What systems did you use for voice coding? I may need to use them soon for similar reasons.

I tried a few. Caster and talon were the best. I’d recommend talon with dragon as it’s speech engine, but apparently talon has a new built in speech engine which people like, so maybe try that engine first to save money.

I'm not a full data point to work from, but I've got neuropathy and nerve damage. Side effects include severe short term memory problems (I can forget some things as quickly as 10 seconds, as frustrating as that is..) and sometimes motor control issues.

For me, speech recognition and assistants, along with software like Todoist, are able to keep me far more functional than I would be otherwise.

> this is the ultimate Rube Goldberg machine

Salty, but I think I agree.

It seems that with AI/ML, choosing/defining your problem well is hugely important. Text to speech, even computer generated natural language is a definite enough task that engineers (and machines) have the feedback to make progress.

Siri is brain-damaged. I am constantly disappointed with how stupid she is compared to Alexa and the gap is widening. I try to keep things as simple as possible with her because she always disappoints me. I wish Apple would open up a tiny bit of those 200 billion in cash reserves on improving Siri.

You clearly haven't had the pleasure of trying to use voice control in a Mercedes-Benz. It makes Siri and Cortana seem telepathic by comparison.

Same with the Audi. I don’t understand why all the car companies aren’t licensing the assistant technology from the big tech companies and using it as the default.

I get Siri via car play, but even car play is sandboxed because it can’t control things like the radio or temperature.

The procurement decisions for the infotainment system are made around 5 years in advance of the vehicle release. The tech is obsolete before the car leaves the dealership.

Because even the "big" tech from "big" companies is still shit. Have a 2020 LG TV with Google assistant built in, and it's a piece of hot garbage.

Example: I frequently switch between display profiles to suit what I need. Saying

"switch to cinema display mode" - works fine.

Saying:

"switch to user display mode" - 100% of the time results in the TV replying "which user would you like to select?".

Like...it's not my fault that the dumb TV has the custom profile named "User".

Google probably spent billions on voice recognition, but it's all worthless, because someone without an ounce of imagination just coded it to react to the words "change" and "user" as the user profile selector, the rest of the sentence be damned.

But back to cars - Google Assistant in Android Auto is equally shit. Try saying "hey google, open spotify", then "hey google, play music". 100% of the time, it switches from Spotify to Google Music. It's insane.

For the reason it took BMW so long to say yes to Android Auto - Amazon/Google will happily license the tech but want ALL the customer data off the car.

Android Auto at least works, and that in turn has voice recognition that works. It's completely "meta" that I have to use voice recognition on my phone to call a number that triggers bluetooth that activates the hands-free and speaker in the car.

If manufacturers use big tech voice assistance, it'd mean extra subscription for the car owners...really sick and tired of these "subscription" thingy. Microsoft Office used to offer one-time purchase, now it isn't an option. Are we going to have everything subscription eventually?

Have you tried speaking to it in German?

(joking aside, is that voice control developed in the US or in Germany?)

German always works! (For cats at least.)

Seriously - my guess is that they try to perform the speech recognition client side (on the local hardware) and are less agressive on how they collect data for model training.

Unlike Google or some company in China which make thin clients that send everything to a central server where it is much easier to recognize, correct and train.

Got a 2020 volvo. The voice controls are insanely bad. I can 100% reproduce a situation where saying "hey volvo, switch off the seat heating" turns it up to max.

Same in my Toyota. It has explicit instructions which, despite a few honest efforts, I can't seem to follow well enough for the assistant to work with any reliability. It's also painfully slow as it fails to understand, making the tumble into the pit of infotainment despair seem to happen in slow motion. Not sure how it ever hit the market.

Is that the voice control in the MBUX, or in older Mercedes?

In every MB I've been in, in the last decade, the voice recognition could be described as "entertaining".

I mean I can understand how stupid siri is. I find it so irritating that I always keep it disabled. But the point that anything Apple doing bad is for the lack of spending is hilarious.

In fact I'd say Apple is just keeping spending money instead of shutting it down. When IBM Watson AI sank like turd in market it cut funding and let most of team go. I think Apple doing same might be more sensible.

I think Siri is intended to be more goal oriented. I use it extensively, with very little complaint, daily-

-set alarm -next song -set timer -set reminder (e.g. "Siri set reminder 6pm close garage door") -send message to <person>

These comprise 99% of what I would want it to do, and it does it marvelously.

> I find it so irritating that I always keep it disabled.

Ah, if there only was a way to disable it fully. Even when I disable Siri, it still randomly decides to call people when I boil a pot of rice

Lernaut & Haupsie went bankrupt in 99 holding the monopoly of every viable dictation platform and the aftermath of a enormous stock fraud that enabled management to hoover up everything at ludicrous valuations eg IBM Via Voice for something like $4BLN cash deal closed in unbelievable time, provided sufficient obfuscation and destruction to lay the sector to rest for two decades.

Microsoft is making a intervention with things purchase the way I see it.

What they have to do is provide a Linux version or at least O365 interop as good with Edge for Linux as Windows.

Nuance support and products are hopeless - my subscription for $120/yr stopped working with ios14 and the app refused to send password reset emails and then we discovered that no online account management existed and cancelled instead of relying on 8/5/300 telephone queuing.

https://en.m.wikipedia.org/wiki/Lernout_%26_Hauspie

https://en.m.wikipedia.org/wiki/IBM_ViaVoice

Siri is a train wreck. I have a $1200 phone where it takes more time to do “hey Siri FaceTime so-and-so” than to navigate. It just sits and spins.

My favourite:

"Hey Siri, make a Facetime call to Bob Smith." "Which Bob Smith would you like to call? bobsmith@icloud.com or bobsmith@me.com?

How at this point does Siri not have even the most basic of logic to know that it's literally the same person. That's Apple's OWN service.

Do you have multiple contact cards for Bob Smith?

Siri is especially bad when compared to Google Assistant or Amazon Alexa, both of which are light years ahead.

Usual Apple reliance on their fans buying their devices no matter how bad the software is.

The thing that makes my iPhone worse is that it's hard to use Google Assistant (haven't tried Alexa) with my iPhone. I can't make Google Assistant the default voice interface. I can say "hey siri hey google", but honestly I'm not doing that. Google Assistant is decent, but it's not amazing. For any non-trivial task, I still have to do look things up manually.

And while I'm ranting... the thing I hate most of about all of these services is they don't have a profanity mode. In the privacy of my own home I have AN EFFIN' HUGE POTTY MOUTH. I love to curse. I LOVE IT. The fact that I can't get one of these services to talk to me like a middle school student is extremely disappointing. I don't want to be politically correct when I'm talking to a virtual assistant in a private setting. I want to say "Hey goognizzle, are there any good mother f***ing movies opening near me this weekend?" and get Samuel L. Jackson style "Here's the mother f***ing movies opening on mother f***ing Friday at the 5 mother f***ing theaters closest to mother f***ing you."*

Edit: TIL how to faux curse on HN. To type display "f***ing", I need to type out "f\*\*\*ing".

Isn’t it more likely that the quality of the voice assistant isn’t that important to most buyers?

(gonna be downvoted to death) Usually Apple die-hard buyers tend to minimize the importance of software/features that are not sported by an i-device or that doesn't work well there. When the 1st iPhone was released, the lack of native apps was a selling point "we use the web standards". When apps were introduced, they were again a selling point (and still are to the present day).

The day Siri will work equal to/better than other competing systems, it will be used as a main reason for which you should buy an Apple device. The day before that, it will still be something that users actually don't want.

EDIT: and to be a bit more clear that I'm serious with this, it actually makes sense. If a feature doesn't work well, you find ways around it. And if you are used to an ecosystem, maybe you don't even know how well another ecosystem works. And finally, if that ecosystem works for you well enough - or that device - you probably really don't care about that missing feature.

I think that's an effect rather than the cause.

My guess is that Apple management doesn't see voice assistants as the "next big thing" to differentiate themselves from the competition. I think VR/AR is where they are focusing and Siri is on life support.

I think so too. I am pretty sure that data will show it too. Five years ago the NextGen computing platform that will change the world where smart speakers. Now it is almost a niche product.

Anecdata but I've never even bothered to try Siri on mine. I think you're on the money there.

Statement about observed behaviour:

> Siri is especially bad when compared to Google Assistant or Amazon Alexa, both of which are light years ahead.

Followed by a statement about the entire ecosystem, of which only a tiny amount is Siri.

> Usual Apple reliance on their fans buying their devices no matter how bad the software is.

I was actually under the impression that this is because Apple actually cares about privacy (unlike the other two) and so is not hoovering up everything you say to it and sending it to the cloud, but rather trying to do as much inference as possible on your device.

ding ding ding

Apple makes money selling devices to people.

Google makes money selling people to advertisers.

As a result Apple's maps and ML assistant aren't as good as Google's which harvests way more of people's data.

I'm perfectly fine with this trade off. Lots of things iPhones do are even presented at as the data never leaving your phone at keynotes/press releases, and ads. A bullet point notably missing from Android phones. I deliberately use Apple's admittedly worse maps application whenever possible because I know it's not telling Big Brother about me when I know Google maps is.

I’m always surprised when people shit on Siri. I use it all the time for simple commands, and it works pretty reliably.

I’ve never used anything non-Apple nor have I used any non-phone voice assistant, so I’m not in a position to compare, but as long as it calls the people that I want, sets timers and alarms as needed, and routes me to my city of choice, why would I care?

>I’m always surprised when people shit on Siri.

>I’ve never used anything non-Apple nor have I used any non-phone voice assistant

You're coming at it without experience. If you try the google and/or amazon version, you may still find Siri acceptable but I can absolutely guarantee you will not be surprised anymore when people call Siri shitty, because companied to amazon and google, it is.

This is not much to do with the speech recognition itself, but with the AI behind interpreting the commands.

I'm indeed also a bit surprised that this hasn't progressed. It's still not possible to say things like "Turn on my living room lights and the hallway as well". It still feels very scripted where you have to say things exactly the right way and in bite-sized chunks to make it work. The same with Alexa by the way.

It's insanely more frustrating because there is no standard list of commands - I'm perfectly fine saying things in the way the computer needs (what is a command line after all) but there's no reference listing what it is expecting so you just have to do trial and error to find out what works.

And it is different for every skill (for Alexa, anyway).

And a command that worked yesterday may not work today. (Gah!)

I would say it has progressed compared to 10 years ago. Speech technology in general to. I remember when YouTube auto captions didn’t even hit close to what was being said we now they’re way more usable.

I just tried that command with Google Assistant and it worked.

Siri is terrible, but have you tried Google Assistant? I find it very accurate at recognizing my speech and very fast. It feels like a vastly superior product and I definitely find myself using it on my Google devices.

Google Assistant is pretty good. But Google's on-device voice transcription in GBoard (unsure if it's a Pixel exclusive) is borderline magical. The delay between speaking a word and having it appear on the screen is shockingly short.

Oh yeah, the transcription is super good. I'm so sad it is crippled on iOS by requiring some weird app switching to work. Except for that part, it is super faster there too.

Yeah, came in to say this. Google Assistant has consistently gotten better year after year. I'm surprised it understands me sometimes.

I don't know about Siri, but I do believe Google Home has gotten worse. My girlfriend and I mostly use it only for things like turning lights on and off, pausing/resuming the chromecast, setting timers, and asking for the weather.

I've noticed it's been misunderstanding my girlfriend more and more over the past few months (she has a slight accent to her English, but nothing remotely difficult to understand), and lately it's started misunderstanding simple things that I say too. For example if I have Netflix paused and say "hey google, resume", 30% of the time it will give me search results for how to write a résumé instead of unpausing Netflix. I've had the Google Home for a few years now, and that literally has never happened before now. To get it to work reliably, I have to instead say "resume playback" or "resume chromecast".

Youtube since recently (or I just noticed it) has very good auto-CC, their voice recognition even works in noisier videos.

Google's appliance recently decided to stream Dark Piano on The Choice on Tune In when I say "OK Google, stream NPR.". Those things are so far apart I assume someone took a bribe to redirect customers for certain keywords. Lol. (Not really but it was my first thought.)

Google's online advice is whenever the assistant fails to work just retrain the device. I remain skeptical that's the issue because it's not like my way of speaking has changed. Instead I have just switched to a new way to request what I need since reporting the issue via Google's vaunted customer service process seems likely to fail.

> Is it just me or have speech recognition platforms like Siri actually gone backwards in the last few years…

I've used both Siri and Alexa daily for many years, and used to rag on Siri a lot. In my experience Siri has caught up to Alexa on most fronts, and I find them more or less interchangeable my common use cases (home automation, timers/alarms, music, news/weather summaries, etc.).

That's not saying much since the Alexa bar is quite low. But like many Apple products, I'd characterize Siri's improvements as "slow and steady" for almost a decade now.

You have draw a distinction between voice recognition and digital assistants. Nuance is really just the former.

Siri is useless for the most part. I would not depend on it for anything mission critical.

I've been having the same feeling with Google Assistant on Android Auto.

It feels like it does better in edge cases at the expense of the main cases.

Its been weirdly variable- my Apple Watch and the HomePod mini seem much more accurate than my iPhone.

Because that's not what she was designed for.

Siri is way more than a pure voice "assistant".

...so he is using it wrong?

Holding it wrong.

>Microsoft tried to buy TikTok's U.S. operations last year in a deal reportedly valued between $10 billion to $30 billion.

>Reports suggest it's in advanced talks with gaming chat app Discord for a deal worth more than $10 billion.

>A report in February suggested Microsoft was eyeing a takeover of Pinterest, worth $53 billion on the public market. Last September, it bought gaming giant ZeniMax Media for $7.5 billion.

Microsoft is in full yolo mode since all other big tech companies have antitrust lawsuit against them. Microsoft spent its time on the cross in the 1990s and early 2000s now they will acquire anything they can.

The fact that none of these companies are competing with Microsoft also makes it easier. That's not the strategy Google, Facebook etc. normally use.

Maybe they are not competing with them right now but they have aspirations to break into their industry and then compete with them. They figured out it was easier to acquire them than try to build it from the ground up.

Easier to acquire AND easier to divest from if there are any monopoly-related/law-related issues.

ZeniMax competes against XBox Studios or whatever their internal video game studio is called.

Since when was Skyrim a competitor to Microsoft Flight Simulator or Age of Empires?

Microsoft also owns Obsidian, that recently announced Avowed[1] which looks really similar to Skyrim. [1] https://www.youtube.com/watch?v=W3QkO8fy3tg

MS's strategy as of late seems to be buying their way inyo being competitive in every market instead of dominating a few markets. I suppose this gives them a great deal of stability and allows them to develop a massive ecosystem of interconnected products and services. It's still a concerning practice... but not as obviously unethical as the monopolistic behavior displayed by other big tech companies.

This was their strategy since always or since they achieved monopoly in PC OS market which ensured them huge profits which they used to acquire competitors or to break into some industry or niche. They would push acquired product or service to millions of Windows users meaning they had huge distribution channel and scale potential.

Microsoft's first acquisition was in 1987 of Forethought Inc. or developers of what is now Microsoft PowerPoint and they bought them for only $14m. Today PowerPoint as a product and as a brand is worth billions.

Hopefully these acquisitions wont end up like Mixer

TikTok was never a serious purchase. It was a political move to throw up a smoke screen when Trump was making noise about banning them from the US.

This is really, really unfortunate. One of the last on-device speech companies is being bought out and moved into Microsoft's cloud division. Expect nobody to be willing to sell you speech recognition without a cloud subscription now.

Nuance has grown through acquisitions. They basically bought up technologies and slapped a common name on them. But each one retained its one oddities. So you configure one product with INI-style files, another one with XML files, a module within that one is configured with text files using configuration codes... it's a royal pain in the behind.

Hopefully Microsoft will eventually unify these apis and configurations into something coherent.

like they did with Azure? xD

I'm in the on-device space, here's a recent demo of my engine for a niche use case (it also does large vocabulary dictation very well): https://twitter.com/lunixbochs/status/1378159234861264896

I'll take a look! I am still looking for the right speech recognition setup for my needs. I wrote an app that largely expects text input of the spoken command, and I'd ideally like to have hotword detection + speech recognition that can be set up to output the detected command to my own software.

What are you building?

I have been hacking on a home automation controller (that isn't particularly well written, to be honest) for a number of years, and voice is one area I do not want to have to figure out how to write myself. I am a bit concerned about having to heavily integrate with a model that requires I explicitly build sentence patterns in their software, because it'd lock me in pretty heavily to that solution.

(My wife has been using Dragon for dictation heavily lately, so that's a use-case that is intriguing to me as well, especially if today's announcement means the death of the Dragon product line in the near future.)

C+AI makes products that can work offline though (yes it has the words Azure and cloud, no they aren't required):

https://blogs.microsoft.com/ai/with-azure-percept-microsoft-...

The blog claims that it is so devices can continue to work when it's offline, not that it didn't use Azure when it was online, and also that the devices are intended to by deployed via Azure IoT Hub.

Which is to say, the blog doesn't make it sound like I can use this without an Azure subscription, even if it works offline sometimes. Whereas the Microsoft Speech SDK, I could just include the DLL files and run with it.

Not quite "on device" but Microsoft Azure provides Speech Cognitive Services as containers that you can run in your local environment:

https://docs.microsoft.com/en-us/azure/cognitive-services/sp...

I believe Google Cloud Platform has something similar available (not sure about AWS).

I use desktop PCs as my "device", so hardware power isn't really an issue here, but the fact that it's still usage priced would be the big issue.

I agree. Even though acquiring is a good strategy for Microsoft to an extent, it is also a problem to see continued consolidation. How can anyone hope to compete against these giants in the same product segments? They have unlimited cheap capital and the ability to integrate an acquired product into the rest of their products. Then there's the issue of intellectual property and continued building of patent war chests. If one of these companies copies your innovative feature, you will have no recourse to sue them because they'll be sitting on a mountain of random mundane patents. The fact that these companies can make $20B acquisitions itself feels broken from a healthy market perspective.

On the flip side, the open source tech has gotten good enough that you can basically roll your own now.

plenty open-source options are available. Checkout https://github.com/NVIDIA/NeMo

Are you saying this discussion calls for some nuance?

You don't know their roadmap or plans for Nuance yet you attribute motive.

Microsoft's focus here is pretty well known: They've retired the Windows Speech SDK almost entirely in favor of Azure services, which is where the Microsoft Speech team has moved to, and the announcement explicitly stated Microsoft is bringing this under their "Intelligent Cloud" segment.

They claimed two things. That this would be exclusively cloud and that nobody will sell you speech recognition without a subscription. I doubt this is all.

Several years ago I used a Nuance product named Dragon NaturallySpeaking that had speech-to-text capability and adds verbal accessibility features on Windows platforms (eg, say "Open Word", speak your document aloud, then "Close Word")

I had no idea they had enough sales to justify a $20 billion valuation. Though to be fair, Microsoft tends to acquire companies at high price tags (eg, Skype, LinkedIn, Minecraft) compared to eg, Apple's acquisition strategy of smaller technology focused companies (other than Beats headphones) like P.A. Semi and PrimeSense.

EDIT: Other comments say Nuance's patent portfolio may greatly contribute to its valuation.

Bear in mind, Nuance is often the voice technology behind other companies' speech-based products too. Nuance technology originally powered Siri, for example.

Nuance itself was actually spun out of an organization called SRI (note the resemblance to the name Siri)

This is incorrect. Nuance was a company that originated decades ago, SRI did build Siri, using Nuance's speech technology, but Nuance itself did not spin out of SRI.

Is Nuance into healthcare or speech/voice recognition? Didn't knew about the Siri thing though

Nuance is a speech recognition company, but due to their on-premise software which has a lot of integration options and having custom distributions for medical and legal jargon, they tend to be a go-to choice for adding voice to professional highly-regulated industry platforms.

Yes. My wife uses Dragon for her writing her clinical notes. It's apparently pretty popular.

Can you link some source of nuance powering siri, i think SRI was where it was built

SRI built Siri using Nuance for the speech components, and Nuance continued to be used for Siri under Apple: https://techcrunch.com/2011/10/05/apple-siri-nuance/ https://appleinsider.com/articles/13/05/30/nuance-confirms-i...

Keep in mind that the company named Nuance is a big, 30 year old public company with a lot of products, which acquired Nuance and changed its name in 2005. According to Wikipedia they did $2 Billion USD in revenue in 2016 and had $5.7 Billion in assets.

Nuance OEMs and sells backend systems for IVRs and all sorts of products. You probably interact with their stuff routinely and not even know it.

They also bought up a lot of small companies and products from bigger companies and own a lot of patents.

Yeah, but they've got their tentacles into a lot of complicated sticky markets.

For example, they've done all the work to get that same software certified healthcare grade and convinced lots of hospitals to adopt it for their doctors. They can sell that for a lot more than they sell the essentially same software to you.

If you go to a corporate website and see an "Industries" section with Healthcare, Telecommunications, Finance, Government, and more you know they're good at this sort of rent-seeking.

Also wasn't Siri based on Nuance technology?

If so imagine them sales...

When you're talking about Siri; you need to talk about the virtual assistant piece separate from the speech recognition piece.

When Siri's ancestor was being worked on at SRI, it used a cousin of Nuance for the speech part. The PI of the virtual assistant piece of that program spun it out as a separate company, and was immediately snarfed up by Apple.

At that time, they switched speech recognition providers; but I don't know if Apple picked Nuance specifically.

Can you link some source of it, can't find it

Like this? https://appleinsider.com/articles/21/04/11/microsoft-in-talk...

I think the way carmakers are going the auto industry could justify most of the valuation.

Nuance spun out their automotive business into Cerence in 2019, so they presumably have zero business with carmakers right now. There should be conditions not to compete with Cerence in the automotive market.

FWIW the acquisition was only 25ish percent above market value. In the public markets Nuance was worth something like $15 billion.

So Microsoft can see your private repositories, your documents if you use Office, your personal files (one drive), your company communication (teams), now they will be able to listen what you talk about. Where is this going? Should we trust this company so much? I think companies like this should be split up.

Years ago, deep in the fine-print of the Skype ToU, was buried a clause in which anything you discussed via Skype granted to MS a worldwide, royalty-free, sublicensable, etc. license to anything mentioned. Is that still there? (Haven't used Skype in many a long year, myself.) I always wondered why that clause didn't render Skype unusable for any business or prospective business conversation.

No, that is standard language that gives a website the right to store and display content you post to it (e.g., to show your avatar on a profile page or show forum posts or chats.) Many websites use it; see this search for examples: https://www.google.com/search?hl=en&q=%22any%20time%20you%20...

This is the text from Skype’s terms: “Notwithstanding any rights or obligations governed by the Additional Terms (as defined below) if, at any time you choose to upload or post User Submissions to the Skype Websites or through the Software (excluding Reports and excluding the content of your communications) you automatically grant Skype a non-exclusive, worldwide, irrevocable, royalty-free, perpetual, sub-licensable and transferable license of all rights to use, edit, modify, include, incorporate, adapt, record, publicly perform, display, transmit and reproduce the User Submissions including, without limitation, all trade marks associated therewith, in connection with the Skype Websites and Skype’s Software and Products including for the purpose of promoting or redistributing part or all of the Skype Websites and/or the Software or Products, in any and all media now known or hereafter devised. You also hereby grant each user of the Skype Website and/or Skype’s Software or Products a non-exclusive license to access your User Submission through the Skype Website and/or Software or Products and to use, copy, distribute, prepare derivative works of, display, perform and transmit such User Submissions solely as permitted through the functionality of the Skype Websites and/or Software or Products and pursuant to these Terms of Use. In addition, you waive any so-called “moral rights” in and to the User Submissions, to the extent permitted by applicable law.”

So, those are both necessary terms due to the US asinine idea that copying data within a system is covered by copyrights, and general enough terms that allow MS to do anything they want with any content you send through Skype.

I don't see how the GP is incorrect. But I also don't see how MS could improve anything here.

I find that very hard to believe.

There was 2 Skypes. Skype and Skype for business.

I can believe this ToU for the former.

It’s there for OneDrive still.

And the worst part, the deal is likely going to be approved (or they would not be announcing it). There's just _no way in hell_ the current political landscape allows "splitting up" a company when it is still allowing to grow it even further.

They also want a piece of Discord, so that's most private group conversations along with the voice chats.

Well, they have been doing that for ages already, remember they hold a monopoly on the personal computer operating system software. Of course now it isn't as easy as it used to be as they couldn't just force IE and Bing into everybody's phone, but they're working on it.

I agree, not that different from Google either.

One difference I've found between MS and Google: MS is much more annoying about everything they do. (Hear me out)

I'd be surprised if there is any real difference in how either company respects your privacy (neither does). But for the most part, google seems to do its thing in the background and leave you alone. Microsoft is constantly popping things up asking you to rate this or that, flinging things you didn't ask for onto your screen, basically the spirit of clippy reborn into modern data collection practices. If I'm going to have my information exploited no matter what I do, I'd rather at least have it happen unobtrusively.

(To be fair, this is just my perception. I was in a g-suite shop for three years and now onto office 365 and I feel like it bothers me about stuff way more often the g-suite ever did)

Microsoft is annoying while Google is creepy. I’m annoyed by the dark pattern designs used by Microsoft, but Google will remember things that you about yourself that you didn’t at moments when you didn’t ask for help. It’s creepy because you’d rather not know that a company has that amount of data on your life.

But Microsoft won't see as much scrutiny as any other tech giant because they are so deeply ingrained in the government's world.

And when they are scrutinized, that most probably puts the scrutinizers themselves in a difficult position.

The report in which the organization explains why Microsofts products are privacy invading, monopolized or otherwise breaching some law or regulation, that report itself is most probably written on those Microsoft products.

Or Amazon

Well amazon can't be neither compared in this, is by far way more evil. At least Google and MS keep some key products open source, so most of them can be checked and re-packaged if there are privacy concerns.

I'm just curious, do you and the people who agree with you think that Apple should also be split up?

I personally am more Microsoft friendly, simply for the reason that they always had and have an open platform to everyone.

I'm honestly interested in people's experiences and views on this topic.

They did not always have an open platform. Windows Phone was was a Apple style walled garden and Windows 8 did not allow applications outside the windows store to use the "modern" user interface.

The only reason Microsoft is trending back to "open" is they failed when they imitated Apple.

An open platform to everyone...

Apparently you missed a few decades of their development of the "Embrace, Extend, Extinguish" tactics.

As well as their tendency to throw exorbitant licensing conditions on nearly everything they sold.

As well as their repeated backdoor deals and maneuvers to cause vendor lock in across the entire PC market.

Nuance is text to speech in hospitals, government, call centers etc.. This is a market access play.

Buy the stock and enjoy the ride.

So you've seen the move "The Circle"? Complete transparency for everyone then?

Personally I think that given current policies an "Umbrella Corp" that does everything is unavoidable. The question is, will I be able to profit through it?

This is a duplicate of https://news.ycombinator.com/item?id=26774367

Note: everyone is getting this acquisition wrong. It's not about Dragon and their Speech-to-Tex. This is all about owning an enterprise Communication Platform (e.g. Contact Center, etc).

I'll repost my previous comment from the other HN thread below:

----

Everyone is wanting to take on Twilio. Last September, Microsoft first announced their Communication Cloud.

A huge focus at Twilio now is moving upstream to the Call Center, where Nuance is a significant player. So Microsoft picking up Nuance makes sense.

It’s clear Microsoft sees communication services as a strategic core part of their business.

(Even at the consumer / gamer level with the rumored Discord acquisition talks)

https://techcrunch.com/2020/09/22/microsoft-challenges-twili...

> Everyone is wanting to take on Twilio.

Well then we can expect Microsoft to swoop in for Twilio any year now.

Given that TWLO market cap is $62B at the moment I wrote this comment, and Microsoft would have to pay a premium over that current valuation, I doubt they have the stomach to purchase someone for $70+ billion.

And given that Jeff Lawson, is both he founder and CEO of TWLO - it's not entirely clear he would sell (unless the number is just so high, he has the fiduciary requirement that he has too - in which case I assume that would be north of $100B+)

I am likely very dense but —- at what point does a (public) company have the fudiciary requirement to sell the company? I mean, I believe you, but I’m curious when that becomes the requirement :)

I'd presume $2T in MarketCap could buy you multiple Twilios even at $100B.

MSFT isn't shy from making large purchases unlike Amazon and Apple: They take such massive bets comparatively frequently.

GitHub at 7.5 billion around 2019 looks like a steal.. weird other software giants didn't try for it..

100% they tried. Can‘t imagine Google and Amazon not wanting GitHub.

But with regards to big acquisitions, MS is ironically and quite surprisingly able to fly under the radar in terms of antitrust. Literally every other tech company is not.

I'd guess its a bit less irony and more that by virtue of having already been through various high-profile antitrust investigations, MS is somewhat uniquely well positioned to understand how to avoid/manage similar scrutiny in the future.

Microsoft can see a lot of data which means they can have plenty of dirt on people involved in anti trust stuff. Beyond certain point companies can do what they want.

A lot of the giant companies are really concerned about anti-trust. Microsoft is under pressure too, but there would be an uproar if Google or Apple bought given their pressure around the app store. Having access to the code that writes the apps would be tough to pull off...

true...Google does have similar product which surely could have caused more anti trust uproar..MS in this space luckily doesn't have such issues..

True GitHub is where the community is.

I am surprised at the amount of mention Dragon is getting here. It has been a long, and at times tragic road. I worked at a company that used Dragon's technology in a voice control product for the original Macintosh. Mostly, I worked on a Windows version for a multimedia architecture Intel was developing, which entailed the acquisition of Spectron for a DSP OS.

I got to meet a lot of the people involved at other companies in the project, including the Jim and Janet Baker, who founded Dragon, and many people at Intel up to and including Andy Grove. It was remarkable that he took interest in what was a relatively small project that was also distant from Intel's core products. I also met Jo Lernout, the L in L&H, which played a role in subsequent tragic events for the Bakers.

All those people, and most of the technologies from those days, are gone now. Dragon ended up a part of Nuance, which itself had been called ScanSoft, and, before that, was a part of Xerox that, if I recall correctly, was acquired by Xerox from Ray Kurzweil.

ScanSoft became a roll-up of a large number of speech technology companies. One of them was Nuance, and the roll-up was rebranded Nuance. Another acquisition was L&H, which had collapsed due to a financial scandal, which blew up after L&H had acquired Dragon. The Bakers got screwed and sued Goldman, who did the L&H deal. They lost.

And that is your capsule history of Nuance. Sorry to give short shrift to the acquired companies I have no firsthand knowledge of.

I believe the real story of Microsoft buying Nuance is that Nuance owns an enormous number of patents.

Oh this is big, Nuance is not very well known to the public but they are indeed very big in speech recognition. We used their solution with automated support systems.

True, i came to know people appreciating it here that it is better than siri,alexa in terms of speech recognition, is it really that good. Support system you mean IVR?

Speech recognition in consumer electronics continues to be slow, unreliable, and a privacy nightmare. I can't tell you the last time I saw someone use it seriously. There are reasonable applications for it, but they are few and far between. I wonder what makes it worth $20B to MS.

Top comment as I'm writing this says Nuance has a strong presence in the healthcare device market, but I'd be surprised if that alone was worth the purchase price.

Talking to Hololens?

I think GitHub got better since the acquisition.

And LinkedIn is still terrible, but no more so than they were before the acquisition. Then again Skype just got worse and worse and Wunderlist just got absorbed by the borg cube. So a mixed record, but at least seems to have been getting better in more recent times.

Wunderlist was killed off before Microsoft ToDo achieved feature parity. The web version wouldn't reliably let me log in. After a few months I switched to Todoist and never looked back.

I must disagree. From a UX side alone, they've regressed, but then there have been more outages since the acquisition than before.

Got any specifics to back up that feeling?

Free private repos? Dark themes? The ability to use your Github account to log into other Microsoft places?

Why are you asking me?

Or much worse: https://news.ycombinator.com/item?id=26667101

Isn’t GitHub still in AWS though? Seems like they’ve just been pushing buggier software lately, perhaps that’s an effect of being under Microsoft but it also might just be an effect of growing faster.

What does Microsoft get out of this? They already have TTS and deep learning transcription, what technical capabilities does Nuance have that they don't have already (or can't develop for substantially less than $20B?)

Probably a crapton of patents for voice recognition.

Also, if you cannot operate a keyboard and must communicate by speech to operate a computer, it's pretty much Dragon NaturallySpeaking or GTFO. Integrating NaturallySpeaking tech into Windows would be a huge boon and further cement Windows as the os to have if you have disabilities.

I have users who have intentionally switched their speech engine from the latest version of Dragon to Talon, for both dictation and commands. Talon is cross platform and directly targets accessibility use cases (far more than just speech input).

I switched from dragon to talon a while back... and then back to dragon :-\ Not to bash though. You've built a great product!

Is it slowly going to turn towards like the OS in movie 'Her', how much of it is really possible?

Nuance are absolutely miles ahead of the competition the second you're looking any other language than English.

I don't think this is accurate (and I work in this field).

And what do you suggest is better? I've worked with nearly every tool (open source and closed) under the sun in medical, industrial, and personal settings and Dragon NaturallySpeaking/Professional was by far the best in terms of accuracy regardless of prosody, accent, background noise, technical terms used, etc.

Personally I think they should've been acquired a decade ago.

Nuance owns a ton of patents and are extremely litigious.

I'm not super deep into the world to NLP, but from what I have kept up with, it seems that the state-of-the-art is almost constantly open-sourced. I understand certain companies have personal relationships already built which may be what MS is after more so than the IP, but isn't there a "canonical" way of building a, for example, speech to text system for your phone?

Is there really a lot of NLP IP hidden behind corporate walls at this point? I just assumed Alexa, Siri, Google Assistant, etc were all using the same model architectures. Genuine question, can anyone shed some light?

I recently went through a ton of speech to text engines to test them. There are a lot of open source research projects as well as paid products, including amazon and google's paid products, which I assumed would be paying for the best of the best. Dragon (nuance's products) blew them out of the water in my experience. I was very surprised. Sentence error rate is still pretty shit across the board, so tiny improvements still make a massive difference in usability.

There is only one English "conversational" voice that I see ( which seems to be the best).

Is there a way to make the others "conversational" if not, is Zoe the best example?

Did you find any competing product that came close?

Thank you

I’m referring to the speech to text product. What’s the Zoe thing?

is Dragon really used in large scale enterprise, unlike siri,alexa for home automation, but i am hearing dragon let alone nuance famous for its AI product for the first time

The hardest part of running a 10 mile marathon for most people is the last mile.

Lots of amazing open source out there but the difference between 90% accuracy and 99.7% accuracy can be very difficult to obtain.

Not to mention, quantified data sets, especially medical ones, can hold immense value.

Absolutely. Take a look at [this list][gov-uk-list] of free alternatives to Dragon. The only good free alternatives they list are the closed source ones built-in to Windows and Mac. Or have a look at [this announcement][nvda-announcement] touting open-source work to integrate closed-source voice recognition engines with an otherwise open-source piece of accessibility software.

gov-uk-list: https://accessibility.blog.gov.uk/2018/09/27/assistive-techn...

nvda-announcement: https://www.afb.org/aw/19/4/15104

a little over a decade ago i contracted for nuance working on a speech-to-text project for cell phone voicemail, either att or verizon. the job was literally listening to actual customers' voicemails and transcribing them (or correcting auto-transcriptions) into the system. literally 6 to 8 hour shifts of pure mind numbing work. we were told nothing about the tech behind it but I wonder if any of that work compounded into some of their in-use tech today, or if it was all just throw away.

pointless story aside, their enterprise valuation at the time was a little under 20% of what it is today, and the company remained mostly the same until 2 years ago. wonder what the catalyst was for their 5x valuation growth?

Kind of data cleaning , any idea what the tech was?

This is incredible for a company like Nuance.

At the time of acquisition:

Nuance: $20B, employee strength 6000. 3M/employee

LinkedIn: 26B, employee strength ~12000. 2M/employee

LinkedIn was much more larger, more 'visible' and perhaps better talent attractor than Nuance at the time of acquisition.

I interviewed there about 10 years ago. An interviewer actually asked me "how will you deal with constantly going to pointless meetings and politics ruining any hope of actually getting anything meaningful done here?" Lol.

Whenever I think of Dragon I always remember the early versions (90s?) I used to have to use that required 1 or 2 hours of speech training.

It felt weird reading 'Moby Dick' to a computer for an hour , but at the time there was no product that could even get near Dragon in any performance metric.

Since natural language interpretation is such a widely used concept now I had taken it for granted that Dragon even existed anymore.

I don't know much about the company, and I don't know much about the product anymore, but that kind of persistence and consistency in this field is admirable.

$NUAN's spike from $45 on Friday makes sense. However, I'm having trouble figuring out why they currently trade at ~$53 given that the market knows that Microsoft will pay $56 for each share. (https://news.microsoft.com/2021/04/12/microsoft-accelerates-...)

Any ideas on why that is?

Pricing in a small chance that the acquisition won't go through (maybe due to FTC clearance, but could be other due diligence related items).

Hopefully they'll fix Nuance's terrible website and support.

Hopefully they will fix a couple of bugs in Dragon as well. I really would like them to have a release focused on bug fixes rather than greater accuracy

More than a couple. It's the most unstable and unreliable program I use on a regular basis.

Are the bugs ML-related or with the app more generally? I've never used Dragon but I'm pretty interested in the TTS space.

Feels like the app, you often have to restart it, highly variable (by orders of magnitude) latency, sometimes won't start at all

If they got around to improving the python hooks too, that'd be pretty fantastic.

I knew Nuance sounded very familiar, but when I googled it, the news were about an AI company, which confused me.

DragonDictate / Dragon NaturallySpeaking, that's why it sounded so familiar.

One thing I considered doing long ago was creating middleware to use NaturallySpeaking via an API and noticed in their license agreement they have the following:

   A license for the Software
   Package does not allow Licensee to use the Software 
   Package on a
   server.

I'm curious how well this would hold up in a court. I only mention this because Nuance is a pretty litigious organization. I imagine they were bought pretty much for the patents.

That seems quite common, I looked at a few PDF products a couple of years ago and they all wanted to license server installations individually whereas for workstations they are happy to offer a redistribution license.

We only buy Nuance Dragon Naturally speaking for our lawyers. It’s the best product out there. So maybe they want the Dragon technology.

In other news, how did Pinterest go from $11B to $53 billion in roughly the last year? COVID-19 bump or something else?

They really turned on monetization and the entire market was discounted 33% last year.

Pinterest went from ~$250m revenue / quarter to tripling that in just 2 quarters. That's huge and unexpected that revenue would climb that high.

What will this mean for Apple? Aside for Siri, Nuance voice is used for the Mac TTS engine, which I use quite heavily.

In English, there are quite a few contenders for TTS, eg. Amazon. Apple can find another vendor. But in some languages it is Google voice or Nuance and there are no other games in town.

Can I just say here that these all-cash transactions really suck?

Here you have been carefully exercising your stock options every month so you can pay only long-term capital gains taxes, and avoid AMT, and then BAM! sorry, you just sold it all at once. "BZZZT. Sucker."

Am I the only doc who despises Nuance? Dragon is definitely Microsoft-style software: lots of exposed features with defaults randomly set to "what no one would want, ever". Nevermind the core algorithm never really seemed to work well for me.

For those wondering some of Nuance strengths in speech: Multiple award winning Very domain oriented and not just general speech Supports many languages

They have also had a boost in stock with covid and more people using remote speech services

What does an all cash deal of $20bn look like? I'm assuming it's really just an ACH transaction or something? There's not a convoy of armored cars with silver briefcases or anything... Is there?

It means that Microsoft are giving money in exchange for shares of Nuance instead of giving Microsoft Stock in exchange for Nuance Stock.

After having tried Microsoft's automatic captioning on Microsoft Stream which is akin in accuracy to monkeys typing something random on a typewriter, I think this seems like a good choice.

Live captioning in Microsoft Teams isn't terrible.

I am confused how speech recognition tech is worth 20 billion. It’s not even on the most popular platforms like iPhone, android or the desktops even. Can someone shed some light on this?

"Enterprise sales" and "health care": https://investors.nuance.com/download/NUAN+%28Nuance+Communi...

> September 15, 2005 — ScanSoft acquired and merged with Nuance Communications, of Menlo Park, California, for $221 million.

> October 18, 2005 — the company changed its name to Nuance Communications, Inc.

So Nuance was worth ~$220MM at the time of acquisition. I'm sure it's worth a lot more now but ScanSoft acquired like 50 other companies too, including old voice recording giant Dictaphone.

Is it too early for me to be hoping that Microsoft will integrate Dragon into Windows and make the voice recognition actually work well?

Off topic but axios looks neat and tidy! I could read news every day. Is there anything similar that covers worldwide news?

In the next decade, there will only be 3-4 companies that own everything tech. We will only have the illusion of choice

I remember using Dragon NaturallySpeaking on a windows98 machine - it was considered pretty good even back then..

People tell me I'm blunt sometimes but have you tried to price nuance lately? Doesn't come cheap.

Now Nuance.

It seems Microsoft is still actively continuing to buy more companies and it seems Discord is still on the menu.

It's actually kinda scary to see Microsoft stretching its tentacles like this. Only this year they bought Bethesda, Nuance and there are rumours about them buying Discord and other gaming studios.

Microsoft's influence is growing deeply...

Where is the antitrust lawsuit?

Microsoft got done for this before (bundling Internet Explorer with Windows) and if you look at where they are now buying out businesses and using dark patterns in Windows 10 to make people use Edge as well as implying it is a fundamental part of the OS they are way beyond was acceptable years ago.

You're right.

They've been abusing their position and power for years now and the Edge situation with Windows 10 is getting much worse after the Chromium update. They force you to use the said browser with some features, like when you use the search function it automatically opens Edge browser, even if you have another browser set as default. What's next? Removing the option to set defaults for browsers?

It's one of the reasons I'm willing to buy a MacBook next...

They're looking at ~$50B / yr in profit, only about ~$17B that's paid out in dividends. What else are they going to do with $33B every year?

And Ubuntu! Totally expecting that in the news at this point.

Really hope not, Ubuntu is like a beacon in the Linux world, no matter if you like it or not. Microsoft purchasing Ubuntu would mean the end of Linux desktop...

I don't know, I wouldn't welcome it either that's for sure. But the github acquisition went pretty OK so far. I don't think it will be disastrous.

I'm referring to the various overtures between MS and Canonical lately, a lot of collaboration going on. https://pulse.microsoft.com/en/transform-en/na/fa2-canonical...

I don't think it would be the end of desktop Linux but it would mean a focus away from the core principles of Linux (which Canonical was not really that good at following anyway).

I was about to say RHEL would be more in their wheelhouse, but then I remembered that ship has already sailed.

I'm shocked at the value of Pinterest (53 billon?).

> The Burlington, Massachusetts-based company, for example, powered the speech recognition engine behind Apple's voice assistant, Siri.

That is the last speech recognition engine I would ever want to buy.

Great buy for MSFT. Voice is the future of healthcare and they just bought the best player in the game.

Rip.

FWIW, I thought Google had pouched all of Nuance's key technical staff years ago.

Jesus, can't companies just be the company they are y'know.

Plenty of blind people would get behind this if it happened. https://t.co/vQIch9PLiW The skype thing was a clusterfuck. I loved skype casts and they fucked it all. Discorde is OK now, but dam does it make me feel old. I join a server and join a VC and here a bunch of 13 year olds. I have to leave because I am 38 and it's creepy to be talking to 13 14 yearolds on the internets at my age lol I made a server for tech nerds and people who like OpenWrt. https://discord.gg/KuNhWzvp5S.

Reply on news.ycombinator.com