Slashdot Mirror


The Future of PC-Audio: Interview With Keith Kowal

red_ed writes "The Noise Report have an interview with Keith Kowal from VIA technologies' Audio division about the future of PC audio. Here's a snippet: 'I think the next big thing will be the widespread adoption of wireless speakers and headphones--cause none of us like a tangle of wires. From a PC infrastructure point of view I can easily see support for these devices being integrated right into the PC.'"

3 of 260 comments (clear)

  1. 32bit audio? No Way by Ironsides · · Score: 4, Informative

    Keith: Certainly for more professional soundcards there is a need for 32-bit support as producers often like to do the mastering in the highest level possible before downcoverting for the final tracks. I agree though that it is more difficult to justify 32-bit support on the desktop. I don't think its really intended as a marketing gimmick, its just more that the HD Audio spec supported 32-bit operation, so they are of course touting that support. I would not expect to see many 32-bit HD Audio codecs in the foreseeable future though.

    Current hadware is only really able to achieve an effective ~20bit resolution due to thermal noise in the components. Unitl we take that down a couple of notches there is no reason to use anything more precise such as 32bit.

    --
    Fly me to the moon Let me sing among those stars Let me see what spring is like On jupiter and mars
    1. Re:32bit audio? No Way by wowbagger · · Score: 4, Informative

      I will have to disagree with you on this.

      When you do filtering, summation, scaling, etc., you frequently find yourself adding multiple values together, then later dividing by some number - for example a filter algorithm would involve adding 256 samples together, then dividing by 256 (roughly speaking...)

      When you do something like that, you need at least 8 bits of headroom in the processing stages or else you either overflow (nasty) or have to drop least significant bits (noisy).

      Running a 32 bit processing channel, even though the final result doesn't need more than 20 bits of resolution allows you to process filters with up to 4096 taps without overflow.

      Sure, you can go to a floating point system, but that adds quite a bit of complexity to the system.

      I work with 24 bit DSPs all the time (Motorola 56301), and frequently have to use the 48 bit long-long format in the core of my routines to prevent overflow. If they were 32 bit DSPs, that would greatly simplify my life.

  2. batteries, boo, small, yay! by claussenvenable · · Score: 4, Informative

    Portable wireless speakers are retarded. I don't need to spend 10 bucks on batteries every few days, and I don't need my shitty portable speakers to have 6 feet of stereo separation between them.

    What, do people think that putting two 5-dollar cost-of-goods speakers 6 feet apart for "big stereo image" makes them sound like anything other than the crap they're made of? Good speakers are large, heavy, and permanently installed with wires hidden.

    There IS, however, a use for this stuff.
    If I need it wireless, it's because I need to move around.
    If I need to move around, it's pretty safe to say I will piss of *everyone* if my music moves around with me (see: bumps in the trunk)
    A wireless set of high-quality in-ear-canal headphones (Etymotics make some, Sony makes some, Apple too now) could be *great*.

    They have very little driver travel, so the power consumption is small, and properly designed you could have a combination headset/headphone setup with passable-not-great quality. It would kick ass to be able to wear headphones under my hat and not have those pesky wires.

    Anyone else use Etymotics and have the painful experience of snagging a wire on a doorknob? Accidentally yanking things out of the ear canal ain't fun.