mark4man wrote on Wed, 27 September 2006 18:45 |
Dan...
The soft saturation function on the AD-122...
...is this similar to "Type IV Conversion" offered up by dbx in a few of their mastering processors? (most notably the Quantum II)...
(ftp://ftp.dbxpro.com/pub/PDFs/WhitePapers/Type%20IV.pdf)
...whereby the top 4db under unity is logarithmically mapped as an "overload region", expanding the dynamic range there by increasing the # of bits that represent the signal. They claim that the high fr.'s are retained as a result of this process, as they are not w/ analog (or typical "look-ahead, brick wall" digital) limiting. (everything below 4dB remains linear.) They also say that the converter cannot be clipped as a result.
Are functions such as these (yours & dbx's) implemented within the A/D itself by means of a digital algorithm...or pre-staged? I remember reading somewhere (maybe Brad's forum) where you said that this can be accomplished in both the hardware & software worlds; and was wondering if; with some of the better plug-in limiters on the market...where they offer so-called "soft clipping", if that is in any way similar to yours (or anywhere near as effective)?
Thanks,
mark4man
|
I do not have the time to examine in detail what dBX did. Their paper is dated 1998, and mine was done in 1995.
At first glance, what they do is fundamentally different then what I do.
To quote you: "They also say that the converter cannot be clipped as a result".
I do not have any such claims. I do my process on the converted data, AFTER the conversion, thus I do not claim that the converter can not be clipped. The idea with mine is to boost the level by up to 1 bit (up to 6dB) for all signals bellow -12dBFS, but that means that the range of 0 to -12dBFS needs to be "squeezed" into 6dB range. So the signal peaks are "flatter", they not clipped, they are "gently pushed down". But if you overdrive the converter, you will clip the signal (flat top).
Such a process is a lot more difficult to accomplish then first meets the eye. It is not a simple matter to get sonically acceptable results, because unlike analog, in the digital world, the harmonics due to non linear processing do not all end up at higher and higher frequencies. In digital, there is aliasing, and the harmonics you wish to eliminate go way up there (from the ear stand point) because the ear is "nearly logarithmic". Of course, there is no way to filter the energy that gets aliased, because they are "manufactured" by non linearity, so by the time you see them, they are already aliased (thus inseparable from the audio).
All those soft limits and saturation processing should be used with care. Some music "wants you to stay linear", other music "likes" the effect.
One factor often discussed by the EE's in the pro audio community is about keeping some margins away from full scale, because some up sampling DA's may "fill in" samples beyond full scale. Unfortunately most high quality studio "monitoring" DA's do not show what many consumers will hear. So I recently tried to "soften" the top 1.5dB on my converters. Some mastering engineers love it, but some did not (I will do it as a "special request"). The reason I throw he above comment in - the digital soft saturation can be helpful in alleviating the rather common practice of "driving beyond the rails" for the sake of "being louder". It is a compromise, and it is difficult to quantify, so I leave that decision to the ears of the mastering engineers.
Regards
Dan Lavry
http://www.lavryengineering.com