Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Input Devices Intel Microsoft Hardware

Skylake Has a Voice DSP and Listens To Your Commands 99

itwbennett writes: Intel's new Skylake processor (like the Core M processor released last year) comes with a built-in digital signal processor (DSP) that will allow you to turn on and control your PC with your voice. Although the feature is not new, what is new is the availability of a voice controlled app to use it: Enter Windows 10 and Cortana. If this sounds familiar, it should, writes Andy Patrizio: 'A few years back when the Xbox One was still in development, word came that Kinect, its motion and audio sensor controller, would be required to use the console and Kinect would always be listening for voice commands to start the console. This caused something of a freak-out among gamers, who feared Microsoft would be listening.'
This discussion has been archived. No new comments can be posted.

Skylake Has a Voice DSP and Listens To Your Commands

Comments Filter:
  • by account_deleted ( 4530225 ) on Monday August 24, 2015 @03:27PM (#50382701)
    Comment removed based on user account deletion
    • Re: (Score:2, Flamebait)

      by s.petry ( 762400 )
      "Objective" does not mean what you think it does. Now go on back to your fantasy about how your opinion is so much better than everyone else and let the grown ups talk.
    • The DSP still requires a microphone to be physically attached in order to "hear" anything, so it should be fairly easy to physically disable -- just put a switch in the microphone circuit, or simply unplug it. Not as easy as putting masking tape over the web cam, but still easy enough for most people.
    • Im sure plenty of slashdotters will invest time and effort in explaining how this can be manipulated by unscrupulous hackers and foreign intelligence agencies to undermine user security.

      Actually I'm more curious as to why this is "new". I could do the same thing with a 20-year-old PC with a sound card (I'm limiting it at 20 years to get some sort of reasonable Pentium with MMX, before that you start to run into CPU horsepower problems depending on what you want to do with the capability), why is it some "feature" of Skylake?

      • The 'new' part is that you can shut down the main CPU entirely and just have a very low-power DSP (and microphone) running. Is put new in quotes, because a number of ARM SoCs have shipped with this feature for a year or two.
  • by LichtSpektren ( 4201985 ) on Monday August 24, 2015 @03:30PM (#50382731)
    Wouldn't it make more sense to have a voice-activated power button on the frame, rather than the CPU doing this?
    • Re: (Score:2, Insightful)

      by Anonymous Coward

      No, because then you could disable it.

  • by Anonymous Coward on Monday August 24, 2015 @03:30PM (#50382733)

    "We made a new processor! It's not any faster but it has an always on mic and exposes a remote-control interface you know nothing about. Oh and did I mention the random generator is biased? You'll love it"

    • >Oh and did I mention the random generator is biased?

      Would you like to substantiate that with evidence?
      I know you can't. You know you can't.
      What is your motivation for saying these things?

      • The problem is that it's very hard to detect, if it is true. Modern pseudo-random number generators use a block cypher and can provide output that will meet all statistical tests that you throw at it, but can be easy to compromise if you know the key. If you were the NSA, then you'd want to have some bits in the key known / predictable, but the rest provided by random electrical fluctuation so that anyone else brute-forcing the key would have to search the entire space, but you'd only have to search a muc
        • You are missing the point.

          AC made an assertion he knows that he or she doesn't know. It was a lie.
          I know it to be a lie because I know the circuit. Several other people on this planet know enough to know it is a lie.
          Other people don't know, which is just how the universe works.

          If you are interested in testing random numbers, you are welcome to buy my book on the subject when I finish writing it in about 20 years.

  • by Anonymous Coward

    Can't imagine a much bigger waste of silicon than this.

    • Wait a minute, which current consumer operating system doesn't have some kind of voice control software? (FYI: It's not the year of the Linux Desktop just yet)

      There's Siri, Google Now and Cortana.

      How is dedicated hardware to decrease power consumption a waste of silicon?

      You need a better imagination.

  • "This caused something of a freak-out among gamers, who feared Microsoft would be listening."

    We're much more realistic these days. Now we understand it's the NSA that will be listening. The nutbars are those innocent creatures naive enough to believe they won't.

  • Oh, I see, this has a special, vibration sensitive electrical switch that you tune to your voice. So when your voice and only your voice says "Computer" (in a thick, fake Scottish accent in memory of the WWII hero James Doohan), will trigger a mechanical switch that lets electricity flow into your computer.

    What? That's not what's going on? You mean the PC is always on in sleep mode, listening to everything thing you say and analyzing it for the words "Computer on", which will take it out of sleep mode?

  • by M.B. ( 4235839 )
    Cool!
  • So touching the mouse or a key is too difficult I guess?
    • by Anonymous Coward
      My hand has become grafted to my penis, you insensitive clod.
    • Push-to-talk is kind of a pain-in-the-ass, yes. Especially if the microphone is across the room, in your entertainment system. Amazon's Echo is selling entirely based on the premise that it is always listening.
      • by GrahamJ ( 241784 )
        If I wanted my entertainment PC to always be listening I'd leave it on, but I suppose there is power to be saved doing it this way.
  • I haven't read anything that indicates the chip does anything other than listen for specific speech patterns and send specific commands to the computer (actually power on is really the only thing I can tell it actually does). I doubt there is any way of retrieving the audio input from the chip. They would have to go out of their way to even incorporate this capability which if it existed would pretty much mean a large number of people wouldn't buy the chip. So until some hacker actually successfully retr
    • Comment removed (Score:4, Insightful)

      by account_deleted ( 4530225 ) on Monday August 24, 2015 @04:39PM (#50383273)
      Comment removed based on user account deletion
      • Because no one makes DSP's on 14nm process, so they'll all use more power.
        Because putting an additional chip in a device makes it bigger.
        Because they probably couldn't even fit an extra 100kb in the space a speech DSP takes up. (L1 and L2 cache aren't small because of die area, L3 is on-die too)

      • by jetkust ( 596906 )
        How does the argument go from Microsoft and the government spying on us to I want to redesign the processor to have more cache and instructions?
      • by DRJlaw ( 946416 )

        The problem is that we have valuable die real-estate being taken up by this shit when additional L1/L2 cache, a core, or other SIMD instruction sets would be better. The market is full of DSP chips, why this, and why on the fucking die!

        Because a separate chip, additional chip-to-chip interconnect, and additional PCB-type lithography is far more expensive than on-die lithography?

        For instance, USB 3 really took off when Intel integrated USB3 into their chipsets. With the 6-series chipsets, USB 3 was a more e

      • by amorsen ( 7485 )

        The problem is that we have valuable die real-estate being taken up by this shit when additional L1/L2 cache, a core, or other SIMD instruction sets would be better.

        L1/L2 caches have hit the maximum sizes you can build before the added latency of larger caches makes the trade-off fail. You can do L3, but the performance benefit is not very impressive for desktop workloads and if you are Intel you do not want your desktop chips eating your server market. Extra cores, same deal: great for server tasks, not for the desktop. SIMD just does not take up significant die space, and the gains are minuscule except for specialized workloads.

        Intel is pretty desperate to find somet

      • by Sloppy ( 14984 )

        Apparently one of the justifications for spending space on such a ridiculously specialized task, is that in the rare event that it's being used, some of the other stuff (e.g. the general-purpose parts) might have a brief opportunity to cool off a bit. Your bigger cache wouldn't have that advantage, because you'd be using it so often.

        Some say often-dark silicon [danluu.com] will be a growing trend.

      • Because the market wants this. Either it'll be a hit, or it will disappear into the idea obscurity bin. You're banking on the latter. Time will tell what the majority who part with money decide.

      • I have two words for you: Dark silicon.

        Since the end of Dennard scaling, the transistor budget for new chips has kept increasing, but the powered transistor budget has not (or, at least, at a much lower rate). More L1 or L2 that needs to be powered all of the time is not easily affordable, but something that only runs when most of the rest of the chip is powered down is basically free (especially something as small as a DSP for voice). Expect to see a lot more of this kind of thing: features that give a

        • Comment removed based on user account deletion
          • Well that just goes contrary to my understanding of what the main CPU is supposed to do, crunch data, and as much of it efficiently as possible in the smallest package available

            Why? Especially in a desktop package, space isn't a constraint. Die area is cheap, heat dissipation is expensive. Your choices are either add some rarely-used coprocessors in the available space, or don't use the space. The cost is the same in both cases.

            Specialized hardware that's rarely used (relatively speaking) should resides outside of it via PCIe bus assuming latency and bandwidth considerations are met

            Latency is one big issue. Another is power. Off-chip communication is slow and very power intensive. The ARM GPUs, for example, compute a hash of each tile before writing it off to the frame buffer, and if the hash is the same as the last time then t

  • I thought we could control our machines with brain waves...

  • I think hardware support for voice recognition would be awesome if it can be leveraged to provide a usefully accurate local recognition capability.

    Yet I very much doubt this will ever happen because the whole point of voice recognition these days seems to be nothing more than an excuse to send data to MS / Apple / Google / Nuance / LEA / whomever.

  • I see a problem with this because of the NSA and their overreach into our private lives. They are already recording our phone calls, they already have access into the internet. So do we really want to have our computer's mic constantly hearing our conversations? MS records what your mic hears for Cortana's "improvement", what stops them from handing it over to the NSA?

    Years ago I'd be considered crazy for this, but now with the NSA being who they the are, can you trust them? Can you trust MS? Ca

  • The "Intel Active Management" (a governor that runs on a secondary CPU independent of the primary one, with cryptographically signed firmware and autonomous access to LAN, WiFi, Memory etc) is also quite disconcerting, but in fact only inclued on certain chipsets (see the tables for Broadwell [wikipedia.org] and Skylake [wikipedia.org]). Unless you are a large institution you probably don't want remote management capabilities.

    It's hard to find which chipsets will feature this DSP but quite possibly some won't. Pay attention when you buy

Our OS who art in CPU, UNIX be thy name. Thy programs run, thy syscalls done, In kernel as it is in user!

Working...