Subject: Re: Re[2]: [linux-audio-dev] 64 bit float or integer
From: Karl MacMillan (karlmac_AT_peabody.jhu.edu)
Date: Sat May 26 2001 - 05:33:56 EEST
I just glanced at this paper, but it looks like it is discussing dedicated
DSP chips. Dedicated DSPs normally make programming with integers
significantly easier for audio than with general purpose processors (by
providing features like hardware saturation, etc). So in that area
something like 48bit integers might make a lot more sense. For general
purpose processors, though, 32 or 64 bit floating-point is the way to go
for sound quality and ease of programming.
Karl
On Fri, 25 May 2001, Rick Burnett wrote:
> I think that a calculation of 32-Bit float to 32-bit integer would
> make more sense. Why go to 48-bit when a power of 2 puts you to
> 64-bit.
>
> Rick
>
> Friday, May 25, 2001, you wrote:
>
> JS> There is interesting papers at
> JS> http://www.sonic.com/sshd_otherinfo.html
>
> JS> Including
> JS> 48-Bit Integer Processing Beats 32-Bit Floating-Point for Professional
> JS> Audio Applications
>
> JS> The master brain behind the Sonic is James A. Moorer who developed
> JS> digital audio workstations at Lucasfilm, 1980 or such.
>
> JS> Regards,
>
> JS> Juhana
>
>
_____________________________________________________
| Karl W. MacMillan |
| Computer Music Department |
| Peabody Institute of the Johns Hopkins University |
| karlmac_AT_peabody.jhu.edu |
| mambo.peabody.jhu.edu/~karlmac |
-----------------------------------------------------
This archive was generated by hypermail 2b28 : Sat May 26 2001 - 07:39:40 EEST