R/E/P Community

R/E/P => R/E/P Archives => Bruno Putzeys (Designer) - Dave Hecht (Master Tech) => Topic started by: Blas on January 02, 2009, 02:33:18 pm

Title: Single precision & double precision regarding AD converters
Post by: Blas on January 02, 2009, 02:33:18 pm
Can someone explain or guide me towards an article as to the meaning of these terms?  I'm under the impression that more A to D converters today are double precision, how does that impact the process?

Joe
Title: Re: Single precision & double precision regarding AD converters
Post by: johnR on January 02, 2009, 05:35:36 pm
Blas wrote on Fri, 02 January 2009 19:33

Can someone explain or guide me towards an article as to the meaning of these terms?  I'm under the impression that more A to D converters today are double precision, how does that impact the process?


Single and double precision usually refer to floating point numbers, which aren't actually used by standard audio converters. Single precision is normally 32 bit floating point and double is 64 bit. Converters, by contrast, normally use integers.

Native DAWs that use floating point DSP will use single or double precision numbers. The terms can also be applied to fixed point systems like PT, which uses 48 bits (double precision when compared with 24 bits) for calculations.

Gory details here:
http://en.wikipedia.org/wiki/Double_precision
and here:
   http://akmedia.digidesign.com/support/docs/48_Bit_Mixer_2668 8.pdf

Edit: It's possible that some audio converters might use single or double precision calculations for additional DSP (apart from the actual A-D conversion), but I can't think of any specific examples.

Edit 2: I forgot about DXD, which is 32 bit floating point samples at 352.8 kHz, so that's a format where the converter does output single precision samples.