Premium Membership - All Voxengo Plugins at a Fixed One-Time Fee - Click Here For More Info
Summer 2018 Discount at Voxengo
Forums     Plugins     Elephant Elephant 1.1e output level

This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.





Aleksey - perhaps this is another dumb question, but I am having difficulty with output/peak readings when using ElephantHQ with Wavelab 4.0g.  For example, when I set Elephant to input=1.2 and output=-0.2 and play a file, the pan/level meters in Wavelab do not register above -0.43 or thereabouts, both in real time and after rendering.  Also, if I stop play and replay, the pan/level meters can vary widely, reading up to -1.83, with no change in input level!  I also seem to notice that the pan/level peak readings are roughly double the Elephant output setting up to about -0.5 in Elephant.  No other setting changes in Elephant, Shape, RShape, etc., affect any of these readings.  By contrast, the pan/level meters always match exactly the output settings of the Waves L2.  Am I missing something (again)?  Thanks for the help.


Actually, Elephant start to limit slightly early than at 0 dB level.  So, in most cases you have to set out gain to 0.2 or 0.4 dB.  Out gain is not a ceiling-type control like Waves L2.  Out gain adjusts output volume after limiting already took place.

Hi Aleksey,

would it be possible to add a ceiling like in Waves L2?  Its important cause the peaks that elephant brings out seem to vary, but i need accurate values like -0.2db and -0.3db etc. to be the MAX value troughout the song.  Clients request that and its important for pressing cd´s as well, as it might come to problems if the max peaks exceed -0.2db. now you might ask, he why not normalize it to -0.2db, laugh now, but i will loose some -0.x db volume if i do that cause the peaks are vary greatly in elephant. i might have like -0.5db in the first 30 seconds and then one peak at -0.1db for a second and the rest runs at -0.5db.  Please make it accurate.


This is not as easy to make as it seems.  The whole good side of the Elephant is the algorithm which does not allow to use ceiling.  If I make it like Waves, I can't guarantee it will sound as good as it sounds now.  So, let's leave this for the future if I ever come to the newer algorithm.

BTW, CD's won't have any problems with peaks.  In any case I highly suspect that before the factory produces glass master for duplication it subtly processes the CD for removal of any possible problems.  I guess (and I've read somewhere about it too) -0.3db 'safe area' is a problem of the past.

Yep, many of the released CDs today, are actually more or less clipped !

A workaround would be, to set Elephant Out Level, to +0,6 or +0,7,

render the audio to a 16bit file

and then reduce the volume of the rendered file, to match the requested ceiling.

Or just put a brickwall limiter behind it and let it only cut off, the last 0.3 db.

Not sure though, if that would degrade the great sound quality, you get from Elephant ...

peace, Jan

Aleksy, you are right.  But the cd should never clip for the sake of correction.  If we ever put out anything like that here, I shudder to see the business drop like a rock

-R :) bert

But with the final 16-bit file you never get clips...  You can have the maximum amplitude, but that does not mean clipping.  You can have clipping only in the 32-bit multitrack environment.  When your 16-bit file is ready, it can't go beyond 0db anyways, and that makes any -0.2 or -0.3 db thresholds the problem of the past.

BTW, post brickwall limiter can damage the reached sonic quality.

Mastermind, you can use Jan's suggestion.  It will work for you I guess, without any major quality loss.  BTW, you can render to a 24-bit file instead of a 16-bit file to preserve the quality during volume change.

Well, it was said, that everything above 2 or 3 consecutive full scale samples, in a 16 bit file is called clipping.

This is still how it is measured.

Problem is that D/A converters, may not produce audible distortions, with 30 or more consecutive samples at fullscale, in many cases.

Some years ago, masters where rejected, if they had more than 3 samples in a row, at full scale.

Mastering engineers than started to fool the industrie with heavily clipped masters, that were reduced by 0.1 db after rendering.

This way no measuring tool would show clipping, although the signal was heavily clipped.

Nowerdays, clipped masters doesn't seem to be a problem, anymore ...

peace, Jan

Hi Aleksey,

i can surely say that labels and cd-replicators still want -0.3 to -0.1db max. maybe some "lower quality" facilities use 0db, but thats not good.

There was a very interesting article on the TCElectronic website (it no longer appears to be there) which was about how (as far as I understood it) due to anti-alias filtering, when converting from digital to analogue, it is quite possible for higher than 0dBfs signals to be produced.  For instance, imagine a square wave -- as the higher harmonics are removed the actual peak value increases (as it approaches a sine wave), so if your square wave was at digital 0dFS, the analogue reconstruction can be higher than 0dBFS.  There is some info about this in RME's help file that comes with their DigiCheck utility -- this level meter actually shows these overs.

I quote from the help file:

"To make this more clear: nearly all CDs currently available show levels higher than 0 dBFS.  Our favourite demo object is Janet Jacksons 'Miss you much' (also on the CD Design of a Decade).  Analyzing this track with a usual Level Meter, -0.2dBFS is the highest level.  So the whole track has been shifted slightly below 0 dBFS, to be sure there will be no problem with the CD manufacturer.

"DIGICheck 4.0 allows you to use both kinds of metering simultaneously, and to easily examine the differences.  IN the second Stereo level Meter, activate OVS.  The scale now changes to a maximum value of +3dBFS, the numerical value is even unlimited.  The track now shows levels up to +3.5dBFS!


"A big part of the discussion about levels higher than 0dBFS covers the audible degradation (lost clearness, rough sound, up to obvious distortion) caused by levels above 0dBFS.  TC Electronics has released a paper covering this topic, which shows that consumer CD-player indeed generate significant distortion when fed with such signals.  The AES paper, which is unfortunately very technically written..."

It is interesting to note (and this could be a marketing point for you Aleksey!), that when I tested a file limited by L2 there were lots of overs indicated by DigiCheck, however when the file was limited to a similar degree by Elephant there were NO overs -- making this a much more DA converter friendly limiting algorithm.

This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.