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

Microsoft buys Nuance for nearly $20B

axios.com

Post reply on HN
Print view View original

301–310 of 418 posts

Re: Microsoft buys Nuance for nearly $20B

#301

Post by radicalbyte »

eghad wrote: post #234
Nimitz14 wrote: post #202

Earlier quoted context omitted.

That answer depends on the language and on your use case. It seems like you're asking about desktop apps, but my parent was not talking in that context. Indeed there's not a lot of choice there because there's no money in it.

I'm even talking vs custom trained models with Kaldi (was working on a startup that was trying to create lessons for public speaking so we could grab enough data to tackle accent remediation/help those with aphasic speech disorders) and again just reiterating, the out of the box performance of Nuance's products are just better than anything else. Obviously Nuance is more than just speech recognition, but still not su…

I've used it in medical in a multi-lingual setting and there it's basically the only game in town.

Re: Microsoft buys Nuance for nearly $20B

#302

Post by geenew »

pie420 wrote: post #282
jxramos wrote: post #182

Earlier quoted context omitted.

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.

No, the real challenge now is how do you minimize the amount of processing power needed, or bandwidth needed, and how do you do all this while minimizing translation time to under 0.5 seconds. Obviously accents and irregularities are also areas I'm sure they are focusing on, but I imagine that optimizing for real time, mobile and low CPU power devices is a huge focus for them.

They ran dragon on contemporary computers, and that was very good at least 10 years ago, probably more. So they have voice recognition working well on what would now be considered very constrained hardware.

Re: Microsoft buys Nuance for nearly $20B

#303

Post by dalbasal »

_delirium wrote: post #235
dalbasal wrote: post #223

Earlier quoted context omitted.

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 e…

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 network…

Yep.

Also, "just give it as much resources as it needs, we're rich" was a game google had already proved willing to win with gmail.

It's easy to lay the tactic out in retrospect. Fund "resource hogs" that users don't pay for. Bet on long term bandwidth costs going down. Bet on major consumer monopolies being valuable, long term. Sounds great and it was great.

OTOH, lets pour $mns into a "business" that we bought for $bns, that has no revenue... because in 15 years we will be worth $trns and it will all sound like peanuts... this was once considered imprudent business planning. Google were willing to do it. Others weren't. Only a few even could.

Re: Microsoft buys Nuance for nearly $20B

#304

Post by mbesto »

perardi wrote: post #268
bombcar wrote: post #190

Earlier quoted context omitted.

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 anyt…

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…

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.

Re: Microsoft buys Nuance for nearly $20B

#305

Post by mbesto »

microtherion wrote: post #281
tambourine_man wrote: post #273

Earlier quoted context omitted.

>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 think it's impossible to accurately determine the ROI of an acquisition with incomplete data. There's way too many factors at play.

Re: Microsoft buys Nuance for nearly $20B

#306

Post by dalbasal »

everdrive wrote: post #124
bombcar wrote: post #7

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.

> 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.

Re: Microsoft buys Nuance for nearly $20B

#307

Post by mbesto »

bombcar wrote: post #190
dalbasal wrote: post #188

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. Googl…

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 anyt…

> 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.

Re: Microsoft buys Nuance for nearly $20B

#308

Post by tambourine_man »

microtherion wrote: post #281
tambourine_man wrote: post #273

Earlier quoted context omitted.

>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 remember people being flabbergasted by the price they paid for PA Semi. Dividing the price by the number of engineers seemed indeed ludicrous. And yet, here we are.

Re: Microsoft buys Nuance for nearly $20B

#309

Post by SkyPuncher »

bombcar wrote: post #7

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”.

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.

Re: Microsoft buys Nuance for nearly $20B

#310

Post by NaturalPhallacy »

fastball wrote: post #98
skrowl wrote: post #40

Earlier quoted context omitted.

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.

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.

Post reply on HN