192khz .. playback fidelity lower than 44.1khz?

Christian Borg christian at borgstation.com
Tue Mar 6 22:47:56 CET 2012


Tony Hardie-Bick skrev 2012-03-06 21:37:
> On 06/03/12 18:29, Jay Vaughan wrote:
>>> "Production increasingly uses 32 bit float, both because it's very
>>> convenient on modern processors, and because it completely eliminates the
>>> possibility of accidental clipping at any point going undiscovered and
>>> ruining a mix."
>>>
>>> the author's obvious lack of knowledge on this and a few other points tends
>>> to undermine the credibility of the article.
>>
>> Really?  Thats not correct?  32bits is a big number space, man ..
>
> yeah - it's a *huge* number space :)
>
> but... it's impossible to dither correctly in FP, so, each time you multiply, you're 
> effectively doubling the quantisation distortion. Keep in mind also, that your mantissa 
> is only 24 bits.
>
> This is maybe okay if your signal only goes through a few multiplies from the input of 
> your app to the output. but, suppose you're running logic. count the multiplies in your 
> signal chain.
>
> so you go to double precision ;)
>
I'm interested but not very good at this (I ask/listen to Dan Lavry and JJ and just flat 
out believe what they say :D) so I have to ask:

Is this a problem with 32bit float (24+8) only and not with fixed 24-bit?

follow up:
I thought many of todays music mixing software used 32-bit float internally because of the 
rounding errors being inaudiable?


-- 
Christian B

  :: "compress that dissonance, if you dare" ::





More information about the music-bar mailing list