Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[amwintx]

At the risk of posting a reply to the original topic   .... I am surprised no one has mentioned the original reason I went to the time and expense of getting an analog mix setup working. It is fun to mix on a real board! I like to turn off the computer screen and  simply listen to the speakers and pull up the faders/set eq/patch in outboard etc. I think it also leads me down different path creatively sometimes. Now back to regularly scheduled the ITB/OTB debate......

[compasspnt]

Both working on an analogue console AND working inside a DAW is fun...Music is Fun...if it's not, you're in the wrong business!

It just comes down to what sounds best, and how much of your music you have to "see"...

[blairl]

zed wrote on Mon, 16 May 2005 05:38

Paul, I have been running some tests in my protools system that have been nothing but inconclusive. Basically what i've done is to grab a session with 20 tracks, mix them as loud as possible without peaking individual or master tracks, bounce, and repeat the above but with trim plugins at -6dB inserted in every track. (they all have some plugins that are time constant) When comparing both stereo mixdowns the difference between them is zero, they basically cancel each other when phased inversely. I'm sure that i'm not understanding the difference between mixing at high levels or low levels since they are giving me similar results! please comment what i'm doing wrong! Best Regards Zed

I think most problems with intersample peaks manifest themselves at the DAC of a system.  If you are internally digitally bouncing and internally trying to phase cancel two digital mixes, then the DAC hasn't yet entered the equation.  If you are experiencing no distortion before your mix hits the DAC then it may be that your original recording without the -6db trim on each channel and subsequent processing level management weren't hot enough to invoke any digital distortion between plug-in processes.

[Paul Frindle]

zed wrote on Mon, 16 May 2005 12:38

Paul, I have been running some tests in my protools system that have been nothing but inconclusive. Basically what i've done is to grab a session with 20 tracks, mix them as loud as possible without peaking individual or master tracks, bounce, and repeat the above but with trim plugins at -6dB inserted in every track. (they all have some plugins that are time constant) When comparing both stereo mixdowns the difference between them is zero, they basically cancel each other when phased inversely. I'm sure that i'm not understanding the difference between mixing at high levels or low levels since they are giving me similar results! please comment what i'm doing wrong! Best Regards Zed

Well, I would guess that your mix and processing isn't producing the peaks even at higher levels - it's a clean mix?  Offending processing normally includes HF EQ boosting, compression/limiting or distortion addition - coupled with instruments producing fierce HF i.e. vocals, small percussion, brass etc.

In this case the peaks may not be seen within the mix itself but may be created in the final product if you bump up the gain and limit hard on the output buss for A&R acceptance?

[Bob Olhsson]

My experience has been that the distortion builds up gradually enough as you work that it's hard to spot before it's gone way over the top. We don't normally listen carefully for distortion after every single change we make to a mix. In fact I'd worry about anybody who did!

The only sensible solution is dropping levels inside the mixer.

Perhaps one reasonable generalization is that with analog consoles, it's often safer to error on the side of being too hot while with digital, it's better to error on the side of the levels being too low.

(And that's not to suggest that all analog gear ought to be run hot, a lot of gear sounds way ballsier with nothing peaking above +10.)

[zed]

Well, the mix i've been using is a fairly standard pop song with no vocals, only drums, bass guitar and electric guitars, the plugins used are EQ and delay, no compression (does compression behaves differently when changing -6dB on signal input even with the same value applied to threshold?*). I repeated the test, this time i didn't even care if the mix sounded good at all, i tried to get the summing of tracks as hot as possible (on the mixdown with no trim plugin on every track the master fader even peaked once!).

Still, when checking the -6dB mixdown with the hot one the only difference between them was that evil peak on the hot mixdown, other than that, they were mathematically identical!

It's truth that i'm not using compression and the added HF on some of the tracks is rather moderate, but shouldn't this be enough to start getting some differences when comparing these mixdowns?

The recordings were all made with apogee converters not going any further than -5dB on input!

Paul, with this tests i'm not trying to prove you wrong, believe me on this, i'm only trying to understand your statement trough practice, for all our best interest it is very important for us to understand our tools as best as we can!

*- i did try this same experience using compression, in this case the difference was very noticeable, although it seemed more like a compression problem...

Best Regards
Zed

[blairl]

zed wrote on Mon, 16 May 2005 10:59

The recordings were all made with apogee converters not going any further than -5dB on input!

That looks like good level practice and may be the reason you aren't able to hear differences.  I think one of the reason's people start to experience the problem is when they record so hot to begin with, regularly peaking at or near 0 dbfs.  If you take a signal peaking at or near 0 dbfs and then try do some plug-in processing, this is where you might begin to see some problems as you try to manipulate the already pegged audiio.

[Paul Frindle]

zed wrote on Mon, 16 May 2005 17:59

Well, the mix i've been using is a fairly standard pop song with no vocals, only drums, bass guitar and electric guitars, the plugins used are EQ and delay, no compression (does compression behaves differently when changing -6dB on signal input even with the same value applied to threshold?*). I repeated the test, this time i didn't even care if the mix sounded good at all, i tried to get the summing of tracks as hot as possible (on the mixdown with no trim plugin on every track the master fader even peaked once!). Still, when checking the -6dB mixdown with the hot one the only difference between them was that evil peak on the hot mixdown, other than that, they were mathematically identical! It's truth that i'm not using compression and the added HF on some of the tracks is rather moderate, but shouldn't this be enough to start getting some differences when comparing these mixdowns? The recordings were all made with apogee converters not going any further than -5dB on input! Paul, with this tests i'm not trying to prove you wrong, believe me on this, i'm only trying to understand your statement trough practice, for all our best interest it is very important for us to understand our tools as best as we can! *- i did try this same experience using compression, in this case the difference was very noticeable, although it seemed more like a compression problem... Best Regards Zed

I understand that your are not trying to dispute anything out of hand and to be honest I am as interested in it as you are. Always willing to learn

I am only quoting from what I have experienced, learned and measured over the years in kit I have designed and the (rather limited) mixing I have done. But still if you are not clipping anything or adding distortion (either intentionally or within plugs) you may still not produce a peak error of the kind I am talking about.
For example, I have a few discs that I take to work for tests cos they are variously clean, punchy or just manically loud and all of them exhibit the problem. One I have of Shania Twain (a particularly quiet track) is possibly the worst of them all exhibiting bursts of up to 3dB reconstruction overs, even though the peak value meters never reach max. Now this disc will (and in fact does) definitely sound different if you try to further process the track, depending on whether it has been re-recorded via an ADC or ripped directly into the system. The reason is that re-recording it via an ADC turns those overs into signal (albeit possibly flattened), whilst ripping the track preserves the original sample values. For instance, if I limit this track under these conditions the limiter behaves quite differently because it's side chain is driven primarily by sample value. It's a good example of programme that is changed by a converter and in fact would still be changed even if the converter were 'perfect'.

[Bob Olhsson]

What's the title? I'm also curious about what happens when it gets lossy coded.

[Paul Frindle]

Bob Olhsson wrote on Mon, 16 May 2005 21:44

What's the title? I'm also curious about what happens when it gets lossy coded.

Its called 'From This Moment On' from the 'Come On Over' album. I haven't ever tested what happens after lossy coding, but my guess it would have similar results unless the decoder clipped internally?

[Paul Frindle]

amwintx wrote on Mon, 16 May 2005 13:09

At the risk of posting a reply to the original topic   .... I am surprised no one has mentioned the original reason I went to the time and expense of getting an analog mix setup working. It is fun to mix on a real board! I like to turn off the computer screen and  simply listen to the speakers and pull up the faders/set eq/patch in outboard etc. I think it also leads me down different path creatively sometimes. Now back to regularly scheduled the ITB/OTB debate......

Very good points There is no doubt that the presence and facility of a good physical mixer allows you to descend more readily into the creative process, as you can interact with it more directly. We should never forget that the product is about creative art more than it is about 'correct' technology

BTW - sorry if my post on possible reasons for sonic differences experienced OTB has diverted this thread somewhat. You are right that all this has indeed been discussed before on previous threads

[bobkatz]

Eric Vincent wrote on Mon, 16 May 2005 01:44

compasspnt wrote on Mon, 16 May 2005 00:03 Very nicely stated, Paul.  Thank you for the awesome detail. Now everyone turn it all down a bit... That means (among other things): Reductive EQing. Yo?

Well, I'm sorry to say that sometimes an EQ reduction can produce overlevels! It sounds counterintuitive, but those intersample peaks are caused by filtering, regardless of whether it's a dip or a boost. Gerzon wrote a paper on this problem.  

So don't count on your "subtractive EQ" to keep the level down if the input level is hot. As Paul pointed out, there are some expensive oversampled peak meters that can measure if you are getting into trouble, but it's a lot safer just to peak to, say, -3 dBFS (max, lower being better) on the final mix and be done with it.

BK

[PaulyD]

Eric Bridenbaker wrote on Sat, 14 May 2005 11:19

Thanks, Paul for clarifying this issue. Recently I've found that running the mixes at lower overall levels has yielded far better results, a certin clarity and punchy transparency that is quite audible, compared to a "hotter" mixing approach. Even though an analog VU meter is too slow to catch these overs, I find it to be a helpful tool, to have some form of metering OTB. I have a question about ITB meters though: Is it not possible to get better metering of these anomalies ITB? There is a plugin by Elemental Audio calle Inspector, for example which is supposed to show clipping instances. http://www.elementalaudio.com/products/inspector/index.html Is something like this of any use in getting a visual on this stuff ITB? Cheers, Eric

Eric, check out this old thread... The link is to my first-ever post to this board.

I have found that the pre-fader metering in Logic Pro does the exact same thing as Inspector. I think most DAWs offer pre-fader metering. It's good stuff. Since I started using pre-fader metering, I have been shocked to find that some plug-ins with absolutely no gain adjustment available will cause clipping all by themselves, no matter the level of the incoming signal. No kidding.

I have to say though, this thread is blowing my mind. I haven't been following it and just started reading it. Just this week I was mixing a project and noticed that pulling the faders down in my DAW really improved the sound. It's kind of funny because in my case it was a revelation born of frustration! I really felt like my eq was good. I'd soloed all my bus reverbs to make sure their darkness/brightness was appropriate for their track. So, I was concentrating on levels to get the mix sounding right. I  would pull a fader down and notice something else now sounded out of balance. So I would pull another fader down and repeat the process. After enough rounds of this, I thought "OMG...Why is this happening?!" Then I thought "Oh well, I'm already frustrated with my confidence in the toilet. And it does sound better. Keep going..." Lo and behold, I finally got a mix I was happy with. And yes, I was surprised to see how low many of the faders were at the end. But I mix through a Folcrom and if the summing is a little soft, it's nothing a little tweak of the makeup gain device can't handle. After reading this board awhile now, I am wary of the ills of overly high digital levels, but I think that's what Paul Frindle was referring to when he said "And to get back to the original subject and my original reason for posting - be aware that by mixing OTB in analogue and encoding to digits via an ADC afterwards - you are removing the possibility of making reconstruction overs in your master. This is very significant within an industry environment where everyone is currently aiming for absolute max loudness and modulation."

I love this forum.

EDIT: And thank you, Paul Frindle. It is truly reassuring to have someone of your technical background affirm that mixing with softer levels is a good thing.

Paul

[blairl]

bobkatz wrote on Mon, 16 May 2005 17:27

As Paul pointed out, there are some expensive oversampled peak meters that can measure if you are getting into trouble...

Here's an oversampled peak meter for Pro Tools designed to detect exatctly what we have been talking about.  It's not too expensive either.

TL MasterMeter - MSRP US$199

[Bob Olhsson]

The trouble is that you can't pop a meter in at every point where there could be an overload which is literally every plug-in slot in a TDM system and even some floating point plug-ins. You also don't want to tie up scads of DSP for metering. Understanding the problem and then watching the levels is by far the best solution.