🚀 go-pugleaf

      I have a few questions about Adobe�s dithering function and what happens
      when AA1.5 opens, converts and processes files. I am working with some
      30- year old piano performances that were originally recorded on reel
      to reel tape and then converted to .aiff format, 16-bit digital at 44.1
      kHz, using Peak LE on MAC OS9. I have �Auto-convert all data to 32-bit
      upon opening� checked in AA1.5 as I need to do some noise reduction and
      equalization. I was initially not pleased with the sound quality and after
      trying several things I unchecked �Dither Transform Results� and �Use
      Symmetric Dithering� and the file sounded MUCH better.




Well, we did Dither 101 the other day here <http://www.adobeforums.com/cgi-bin/webx?13@@.3bbb80b6/6>, and I have to say that if you are talking about an open reel tape with typical tape noise levels, then if you are hearing differences, it's unlikely to be dither causing it, because even if this is a good R2R recording, the dither is going to be buried about 30dB below the tape noise, since it only operates at the digital noise floor, which is -96dB or thereabouts. And if you open this file as a 32-bit one, and do some noise reduction, then you will effectively remove the remaining dither artefacts anyway, by virtue of the NR process.

      Is Adobe is applying dithering when files are opened and when it converts
      from 16-bit to 32-bit?




No - you can't apply dither to a floating point file, unless you very specifically add it at a fixed level in the dynamic range (which is what Ozone does).

If you simply open a file, then you haven't done any transformations, so 'Dither transform results' does nothing - similarly, Symmetric Dithering does nothing whilst the file is in 32-bit mode. Come to that, nothing makes any difference to the way a file opens in 32-bit mode - because upon opening a file, nothing at all initially happens to your 16 bit audio at all, except that 7 zeros are appended to the end of each 16-bit sample, and some empty scaling bits are added. If you are hearing a 'clear difference', then it's going to be for another reason. I'd guess that this is more likely to be with differences in the way that your soundcard handles 16-bit and 24-bit audio (it will be effectively 24-bit integer that's sent internally back to the card), but hey, it may be something completely different. Whatever it is, if it really is 'clear', it's not going to be caused by dither, which only gets added to a file when a signal is converted back to 16-bit. It may be that your soundcard has the option to dither results sent to it - this is quite feasible, although not a normal thing to do, I believe.

      Why does it take AA1.5 four times longer to open a file when �Dither Transform
      Results� and �Use Symmetric Dithering� are unchecked then when they are
      checked?




I tested and timed this - it makes no difference at all on this system, which is exactly as it should be. I suggest that you retry the experiment with the the .pk files disabled - that's the only thing that can possibly make a difference to this - and it does!

      When �Dither Transform Results� and �Use Symmetric Dithering� are checked
      is dithering applied to all effects including noise reduction?




It would be if you were transferring the arithmetic results (32-bit) back to a 16-bit file, yes - but if you are operating in 32-bit FP, no. Dither is only applied if the results are being used in an integer format session.