1. Register NOW and become part of this fantastic knowledge base forum! This message will go away once you have registered.

How do you take 24bit audio down to 16bit

Discussion in 'Recording' started by dabmeister music, Aug 15, 2003.

  1. dabmeister music

    dabmeister music Active Member

    I've read up about taking 24bit audio down to 16bit without loosing the sonic clarity of the original composision. My DAW software has the Apogee UV-22 algorithm included. Is this used or is there some other method? Or is this process considered as truncating or dithering? :w:
     
  2. Kurt Foster

    Kurt Foster Distinguished Member

    UV-22 is an Apogee dithering product.. It takes a 48K digital signal and dithers it down to 44.1

    I come out of my audio card, a Dakota at 24 bits through the s/pdif output and run it straight into my Fostex CDr burner. The extra bits are dropped automaticly. I can also go from my CDr burner at 16 bits into the DAW upsampleing to 24 bits.. This works great and sounds damn good..
     
  3. dabmeister music

    dabmeister music Active Member

    Kurt I have a question , what platform are you working with? I'm currently on WinXP.
     
  4. dabmeister music

    dabmeister music Active Member

    And I'll tell you why I asked. I have the Sony CDR-W33 Standalone Burner and I have the SPDI/F output from my Layla24 going into it. I had no problems transfering digital audio like this when I was running Win98. But all I get now is "den unlocked" or something to that nature. What could this be? It only started when I upgraded to XP.
     
  5. laptoppop

    laptoppop Guest

    Actually, there are two related things here - but its good to not confuse them.

    First is sample *rate* conversion. CDs are sampled at 44,100 samples per second. Recordings are often at 48k, 88.2k, 96k or even 192,000 samples per second. The advantages of the higher sample rates are a huge matter of debate. Some folks say its worth it, others say don't bother - especially over 48k. Software needs to take the data as recorded and convert the sample playback rate to match the CD rate. There are several related ways for resampling - some better, some worse. Cool Edit Pro will let you choose what way to use - the better ones are a lot more computationally intensive and take a lot longer.

    Strange note: Sonar is great program, but does not do sample rate conversion -- one strange omission. That's why I use Sonar *with* Cool Edit Pro for my own work.

    Next is bit depth conversion. Each of the samples at the previous rates has a particular size -- for audio it is typically 16 bit or 24 bit. There are some huge advantages to using 24 bit while recording and mixing, and going to 16 bit after all processing. When you are going from a higher bit depth (like 24-bit) down to a lower bit depth (like 16-bit) you can get there by just chopping off the lowest bits -- a process called truncation. However, a while back, folks discovered that by adding a smidgen of noise (dither) to the signal, it can sound a LOT better. Think of it this way -- if you are trying to represent 4 1/2, having a signal vary between 4 and 5 will be closer than a signal that is stuck at 4 or 5. Apogee came up with a particular way (UV-22) for adding dither noise to a signal that sounded exceptionally good at the time. There are several different algorithms for adding dither-- and each has its uses, and might actually sound better for particular sound material. Cool Edit Pro will let you choose between different dither algorithms.

    http://www.izotope.com/products/audio/ozone/ozoneguide.html has a couple of PDF files that explain this in much better detail, and has some sample audio files to illustrate how it sounds.

    -lee-
     
  6. laptoppop

    laptoppop Guest

    I could be wrong, but I wonder if this is related to the copy protection bits in the spdif interface...

    -lee-
     
  7. Kurt Foster

    Kurt Foster Distinguished Member

    Windows XP Pro.. The copy protection could be what's happening.. I use a Fostex CDr recorder that doesn't have copy protection. Consumer burners do impliment copy protection schemes..
     
  8. Opus2000

    Opus2000 Well-Known Member

    Kurt

    UV22 ONLY DITHERS!!! Does NOT SRC(Sample Rate Convert)!! BIG DIFFERENCE!!!

    Always bounce down from 48 to 44.1 THAN Dither down to 16 bit.

    Opus :D
     
  9. by

    by Guest

    A year or so back there was a nice website that provided a full featured double blind test for comparing different dithering algorithms. I think UV-22 was one of the winners. I can't remember. I remember the Waves dithering plugin was voted higher as well. You could really tell the difference between some of them. some of them had horrible sounding dynamics sub-noise floor.

    Also to consider is noise shaping, which will change it so there's less 'hiss'.
     
  10. by

    by Guest

    oh, here you go.. sorry if i was wronge

    http://www.24-96.net/dither/blind_shootout.htm
     
  11. thedug

    thedug Guest

    So you are saying to export a mix down from 48 -> 44.1 at 24bit. Then take that and bounce down to 16 with dither?

    If you have a 32 bit mix engine couldn't you argue that you'd still need dither for the 24bit mix?

    d./
     

Share This Page