Crysonic SpectraPhy - no oversampling/look ahead limiting?

[Compyfox]

I want to check that with other users of the SpectraPhy, not that it's on my end only...


I recently got an engineering job to "brush up" a bootleg recording from one of the last live sets by Shael Riley and his crew from NY. This recording was probably done with a portable recorder and I couldn't do much about the panning issues, but this is not my main problem.

Of course, being a live recording and all, you also hear the crowd, especially cheering and clapping. Usually that's not a problem. I already turned down the cheering in the pre-mix before I took the work into Cubase for better automation, where even here the applauding audience was once again turned down.

The final step from the 48kHz track was a volume raising, track mark setting for CD and dithering to 44kHz/16bit. My main weapon of choice for this is Wavelab 4.


I used CryQ (roll off from 0Hz to 150Hz of about 1dB, raising subtle the upper frequency ranges from 8kHz to 20kHz of about 1dB max), SpectraPhy (Thresh -8dB, Physical/Digital to 80% digital, hard/soft to about 25%, "tube" to around 10%, limiting to -0,3dB) and InspectorXL for metering (the EA version).

While listening to the complete set, I never got any form of clipping. iXL always showed -0,33dB - no full scale peaks. However while rendering and reanalysing in Wavelab, the peak of tracks with clapping exceeded the -0,3dB limit.


I tried this with Yohng W1, TbT TLs Maximizer and Buzzroom Maxi. All with the same results. I didn't test Kjaerhus or Voxengo as I don't have any of those master limiters, neither Ozone nor L1 from Waves as I don't use them either.

I turned off the EQ too, even exchanged it (parallel to the limiter) - always the same results. Which brings me to one suggestion... all those limiters are not capable of oversampling/don't have a look ahead mode - to catch the spikes. I first thought that the tube is accidentally put "after" the limiting process, but it looks like it's parallel to the attack/release (physical/digital) and knee mode, especially since all other limiters I used showed the same behaviour. So it has something to do with oversampling.


Does that also count to SpectraPhy or is there a look-ahead mode integrated but we didn't find it yet?! Can you (the user) or Crysonic confirm that?

Also... any news on the "native plugin" front? Or has Specra'C and Cry'Vocal once again more priority.

[flugel45]

I don't do any mastering of my own...

But I recall reading something (In this forum) about Wavelab affecting the levels in a different way (Or maybe it was conflict between Wavelab and something else...?).

In any case, I'm sure someone else will come along with the exact answer soon.

(Sorry, I realize this isn't very helpful. I just thought I'd mention it since I vaguely remember reading about Wavelab doing something weird under certain circumstances. Of course, I may be totally wrong about this).

Good luck,
JD

[Compyfox]

Well I guess it has something to do with the spikes from the clapping, as most limiters can't catch them. Unless I use a compressor with very low attack while the in-depth mix, which doesn't work cause I only have stereo files and this'd affect everything.

Well Wavelab never had any real issues, so I guess it's not the audio editor, but the limiter themselves. I remember however that Bob Katz wrote me that limiters who can't do oversampling (which would make it possible for them to limit at 0dB) should use an output setting of -0.3dB for digital spikes to have enough headroom.


The funny thing is, while prelistening in Wavelab and in Cubase I see no clipping with iXL. After rendering and checking the file again in Wavelab (with the analyze function), I get 0dB readouts in the tracks with clapping in it, everything else is at -0.3dB. And no clipping incident.


Maybe the SpectraPhy was not made for this (even though it sounds that good), but that'd be an idea for a v1.5 or something like that... a look ahead mode.

[crysonic]

Hi Compyfox,

Apologies for not responding as I have been interstate for a while on holiday.

SpectraPhy ' is a look-ahead Limiter ' however there is no internal over sampling. The saturation (tube) process is not placed parallel to the limiter, it is the first process when active, otherwise you should see over's with iXL during auditioning.

We will be adding a variable look-ahead mode in the next version to eliminate these problems, however in this case I do not thing this is the cause.

Do you have access to any other host for testing? as we have not seen this before but we are aware of other issues depending on the version of wavelab which will be addressed when the native version is released. Thanks,

Michael

[Compyfox]

Ah don't sweat for being out for a while, I can wait.

Well I didn't test it in SX3 to be honest, and I unfortunately don't have any other tools to access. The funny thing is, not only SpectraPhy does this, all other limiters I tried did the same, it is always the same behaviour. Maybe the audience applauding causes spikes that neither of the limiter can handle, but I get no clipping incidents - that's even more strange.


By new version, do we speak about a V1.x or V2.x? Also I'd feel more confortable if I could know what the attack setting of SpectraPhy is (unless that's really handled with the digital/physical knob). And how far is the "look ahead" setup at the moment (I'm a bit new in this case, though - maybe it's time to do more research on it)?


Furthermore - what's the status of the native version betatesting. I hate to bug you again, but you know... it's always postponed.

Anyway... I can confirm that SpectraPhy and CryQ works in Wavelab 4 - without installing problems like with Spectralive. So does nXtacy and SpectraQ, sans the LED lightings of course - strangely only partially in SX3 too. Though I didn't test either of those two while rendering in Wavelab - still prefer X-Citer and Clas as exciter to be honest, and so far CryQ worked just as good as mastering EQ (unless that "stereo widening/narrowing feature is implemted instead of the phase shifting function).