Academy Dashboard Forum Production Mixing Setting levels Reply To: Setting levels


    Michael Wagener (King's X, Ozzie, Metallica) starts with kick drum set to -20dB for monitoring. The snare should be the same apparent level. Keep in mind that there's a difference between metered level and apparent level. In essence the snare should sound at a level equivalent to the kick regardless of what the meter shows.

    The following comes from an article I wrote for Westlake Pro Audio - the concept was explained by Paul Frindle, I'm basically paraphrasing. If you're not familiar with Paul Frindle, every time you watch a video of Warren in his control room, he's sitting in front of a rather large example of Paul's work. He also was involved in the design of the Sony Oxford digital console and Sony Oxford Plug-ins, now known as Sonnox.

    Thinking inside the box

    The big problem with ITB mixing comes from the illusion created by software meters imitating hardware meters as we listen to playback. The illusion is thinking that you’re dealing with audio signals when you watch the meters during playback, when in reality, you are seeing numbers, which have a skewed correlation to what is actually happening.

    What you are not seeing, is that what looks like an acceptable signal below that which any overload indicator can show, is in reality something that cannot pass even remotely correctly out of your digital mixer at full level. To demonstrate how this may affect sound quality when mixing ITB, it’s strongly recommended that you do the procedure below and see for yourself. Once you do, the concept will go beyond mere words and inform the way you mix ITB. The following example is a worst-case scenario, but it illustrates the problem.

    To see what happens with an apparently “legal” signal, you’ll need a workstation such as Pro Tools, a signal generator plug-in with a good filter section that actually goes flat to 20KHz and rolls off at 24dB/octave or so, and a good quality EQ filter.

    Set up a mono channel in Pro Tools, or whatever DAW you’re using. Insert the signal generator plug-in at the beginning of the channel inserts and set it for a sine wave at 1kHz-2kHz. Follow the signal generator with the EQ filter plug-in and set it for the maximum slope at 20kHz. (For example the Sonnox Oxford EQ plug-in has a has 36dB/8va slope at 20kHz, which illustrates this exercise well (any other good high-frequency filter should work as well.) Set the EQ filter plug-in on bypass.

    To prove that what you see isn’t what you get, set the channel fader at 0dB and make sure the meters in Pro Tools show -6dBr (you may have to adjust the output level of the signal generator plug-in itself). Once your main stereo bus shows -6dBr with channel fader at unity, switch the EQ in and out using the bypass button. You’ll see that there is no effect at the main stereo bus; the output level will remain the same.

    Now, switch the signal generator to white noise and note that the level at the stereo bus is still -6dBr. Next, take the EQ filter plug-in out of bypass and watch the signal level rise dramatically. Wait . . . what!? In the case of the Sonnox Oxford 36dB/8va filter, the meter level will rise a full 5dBr to nearly 0dB, which is the digital ceiling.

    Since EQ is simply a frequency-dependent amplifier, a cut in amplitude in an analog system should show a decrease in output, certainly not an increase. A cut at 20kHz would show a very small dip in output voltage if anything at all, since not much is going on up there to begin with. However, a 36dB cut at 20kHz in a digital system shows an output boost of +5dB. How is this possible, you may ask?

    I'm glad you asked. The answer is to remember that we’re looking at numbers, not voltage. A digital signal generator plug-in produces sine waves correctly, but when switched to a noise setting, it becomes a random number generator that’s driving the output.

    Despite the fact that when set to -6dB peak, no sample ever gets to be greater than 50% modulation, a reconstruction of the undecoded sample values produces nearly full-level signal. (In order to produce a smooth analog signal from the digital input, DACs employ reconstruction filters.) Reconstruction is filtering and therefore, the EQ filter plug-in is acting like a partial reconstruction filter (much like a DAC), which in turn is feeding a more legitimate signal that the sample-value meter can read more correctly (hence the visual increase of 5dB showing up on the stereo bus meters).

    Now, if the -6dBr noise from the signal generator was passed directly through the DAC sans EQ filter and reconstructed correctly (remember that reconstruction means filtering) it would produce nearly a full modulation signal even though no sample gets to be bigger than 50% and no reading says it’s bigger than -6dBr. In essence, the DA converter is reconstructing the white noise with full-level signal at its output, but without any visual evidence that a full-level signal is being passed.

    If your EQ filter is a good one, you should be able to switch it in and out and hear no difference in the sound of the signal from your DAC, despite the DAW meter reading wildly differently. The filter has neither added nor taken anything significant out of the intended audio signal, but you have nearly doubled the sample values within the Pro Tools channel.

    If you boost the signal generator’s level up to -2dB or -3dB (still less than only 75% full level), it now clips when the EQ filter is unbypassed. As such, the sound definitely changes when you switch the filter in and out. Because it’s mathematically limited and in error when the filter is in, it cannot pass through the output of the filter. What’s happening in your DAC is that it’s saturating at the output while reading -3dBr within the mixer itself. The result is an illegal signal with no visual indication that anything is wrong.

    What does this mean for a mix? With cymbal crashes, high-frequency EQ, limiting, and etc mixed together, output signal can be a bit like white noise within an actual production, even if none of the contributing channels hit the red light. This is the exact register we refer to as “air” and resolution. To further compound the issue, engineers are aiming at maximum possible mix output levels on meters that do not show signal.

    ITB mix vs. OTB

    So why does an OTB mix apparently sound better than an ITB mix? If you’re modulating multiple DAC outputs at levels close to 0dBr, all of those DACs (even if flawed), are working to legitimately reconstruct your output channels before you mix them together and produce too many illegal signals. Oddly enough, the loss of sound quality due to all of those converters combined is not as bad as the illegal signals created within the digital mixer by the over-modulated signals trying pass through the DAW’s main stereo bus.

    If numbers are the one thing that computers can add up almost perfectly, what’s the problem with ITB mixing? It’s not a summing issue per se, rather an illegal output problem caused by the fact that there are no meters that display actual signal in your entire mixing environment — you simply never see it happening.

    Now, pull up your favorite test mix on your DAW, re-mix the whole thing making sure that every place in all chains (including between all plug-ins) never gets bigger than -6dBr. Make sure your final output after any limiting etc. also never peaks beyond -6dBr. Now do the comparison between this ITB mix and a similar OTB mix. You may be in for a little surprise.

    Hope this is helpful - and sorry if it's been covered before.


    • This reply was modified 7 years ago by soundsuite.