Cool Edit Quality Setting During Resample?

[neo_74]

Hi, I'm using Cool Edit Pro to re-sample the bit depth of a .wav to 32 bit. I noticed under sample rate there is a slider bar representing quality from low to high and the default is set to 150.
What exactly does this slider bar do and should I adjust it at all?
Also should any of the dithering settings be changed as well?
Thanks for your input.

[Cornucopia]

There are a number of algorithms that could be used when doing resampling, and "orders of magnitude" in precision particularly with regard to high-, low- and band-pass filtering. The better one gets, the longer it takes, but the more accurate the conversion with less amounts of ringing due to the filter's phase characteristics.
Dithering is also important WRT anytime one does processing (including mixing, DSP, etc), or in converting to a format with less bit depth.
Best is probably:
Triangular 1/3LSB (or maybe Gaussian) random dither + Rounding + Truncation
..then Rounding + Truncation
..then just Truncation.

Makes a big difference especially with "musicality" and "ambience" naturalness.

Scott

[neo_74]

Do you keep Dither enabled during bit depth convert or is only to be used on the final .wav file?
I am taking my 44.1kHz/16bit .wavs and change bit-depth to 32 or 64 followed by a convert to 48kHz/16 bit in order to make it dvd compliant. Just wasn't sure if dithering is done throughout the whole process?

[Cornucopia]

Any time you "number crunch" and/or lose bitdepth, dithering is a good idea. This includes "simple" mixing. That being said, you want to make sure it is the right kind and amount of dither, because you could have a "buildup" of random noise (aka white noise hiss) if you used too much (though even then, it's usually not that much).
The tradeoff in the maintenance of quality/naturalness vs. extra process time, possible steps & ?noise buildup? leads some to want to use it ONLY on the final downconvert/downsample/downrez.
I believe, with the right system, it should always be there.
(I use a ProTools system that allows you to specify the inclusion of dither--even 3rd party--to be used on a system-wide level. AND IT WORKS GREAT)

I notice you're following the suggestion about changing the bitdepth BEFORE doing an SRC. That might be needed for some so-so systems, but the better SRC algorithms do that automatically internally (and often with dither, especially when they use floating point math on the calc), so it's a bit redundant.
If it works for you, fine.

Best thing for you to do it TEST a bunch of variations, noting the different settings and steps taken, and find which one you prefer.

Scott

[neo_74]

Hi Scott, I have been using dither enabled throughout the process so far with cool edit pro.

48kHz Noise shaping curve
Triangular p.d.f.
0.5 bits dither depth

But now that you mentioned it I do notice some White Noise Hiss buildup within the track I've been playing with, any suggestions? Seems more apparent after the convert to 48khz.

I'm undecided still if I want to have my music playable through dvd or cd that is why I'm asking all the questions. So far the pros I see are that cd audio has no post processing and less is lost if the cd should someday become a coaster.....

[Cornucopia]

Depends on what you:

1. Started with
2. Are doing in the middle
3. How you're going to be decoding/playing each

Conceivably, BOTH DVD and CD would be, at most, 1 generation down from a studio master. These would then each be stored as LPCM and sent direct (raw digital stream) to the same D-to-A converter in a HiFi receiver and same speakers. Should be the same experience either way (in this scenario). Difference between 44.1kHz and 48kHz is minimal. Only time I've ever been able to tell has been when the Analog LowPass filters for 44.1k, supposedly at 22.050k, are non-linear enough and dip into the normal hearing spectrum (whereas 48k, with 24k filters, might not). My hearing used to be up to ~19k, but I'm in middle age now, so most likely it'll be dropping soon if not already.

Scott