Sample Rate and Bit Depth Conversion in Cool Edit Pro
Posted by Syntrillium on September 02, 1998 at 10:32:42:
In Reply to: sample-rate/bit-depth posted by Peter on September 02, 1998 at 09:51:48:
On getting the highest quality results (best s/n ratio, least noise and artifacting, etc.) here are our suggestions:
1) Working with single wave files, if doing more than one process to a part of the file (eg Equalize, Dynamics, then Amplify) it is best to open the file as 32-bit (you could check the 'open all files as 32-bit' option, but this would cause 16-bit files to always be converted on opening which makes loading slow - you may only want to convert the files you intend to modify in some way, in which case Open As will work just fine, or opening the file like normal then Convert Sample Type to 32-bit).
1a) When done modifying the file, save as 32-bit if you intend to modify it more in the future (or use it in a multitrack mix where it will be mixed with other waves later). If you want to save HD space but still intend to mix or modify the wave later, convert to 16-bit with Dither enabled, Depth 1, No noise shaping, and pdf of Triangular for very low artifacting or Gaussian for no artifacting but slightly lower s/n ratio. If you are not going to ever modify or mix this final waveform (and/or it is going to CD next) convert to 16-bit with Dithering enabled, Dither Depth of 0.7, pdf of Shaped Triangular, and Noise Shaping set to the 44.1KHz setting.
2) If you still are modifying with 16-bit audio (eg if only doing one transform to it), then it is best to have Dither Transform Results enabled to prevent artifacting. The s/n ratio decreases by about 1.5dB (or is it 0.7db?) each operation you perform (with 32-bit audio the noise floor remains the same after operations for s/n ratios below 120dB or so).
3) If editing portions of audio, it is best to have Smooth Edit Boundaries by Crossfading enabled (and set from 2 to 5 ms). If enabled, most transformed sections will be crossfaded with the neighboring audio to prevent 'clicks' at those boundaries. Amplify does not do this, so use a flat Envelope instead set to the amplification you need (or draw your own amplitude curve using the Spline option to attenuate or boost more smoothly).
4) In multitrack, set Mixdowns to 32-bit so when you do a mix down it is highest quality. You can then convert to 16-bit with the dithering options you want. If Mixdowns are set to 16-bit, audio is still mixed as 32-bit, but converted to 16-bit with standard Triangular dither and no noise shaping.
4a) For higher quality playback in multitrack, set Playback Mixing to 32-bit. But, if outputting to several output devices (like one output device per track) it is best to set playback mixing to 16-bit for performance reasons.
That's all I can think of for now.