Thanks, @Agoldnear .
A little bit of knowledge is such a dangerous thing. I upgraded my 20+ year old speakers to some KEF LS50 Meta’s. Pretty exciting. I set them up and started playing Horowitz with noise leveling off. I heard the “clipping” effect on several tracks. I tried your suggestion about software volume control. Lowered the volume by about -8db. “Clipping” effect went away.
The bass on the KEF’s seemed kind of overwhelming. Then I remembered that due to what I perceived to be a poor bass response with the old speakers I had been using the Melda equalizer to boost the bass end of the spectrum.
I turned off Audio Units Effects. KEF’s sound great, and the “clipping” effect went away. I haven’t tried the same tracks with volume leveling yet. I will do this and report the results.
I’m signing off to listen to the KEF’s some more. It’s a totally different sound experience. I feel kinda like the man who fell to earth.
There is direct correlation to the EQ boost settings you are implementing and the clipping distortion, because you are adding digital-gain by virtue of that DSP at those frequencies… So you must compensate at the plug-in, by lowering its output… A single plug-in, before any processing, adds approximately +2dB of signal gain and then on top of this gain, you are adding more gain in the EQ processing when boosting the levels of those frequencies, which adds bits to the fundamental digital-audio signal data… So pay close attention to the level displays of your EQ plug-in so to keep from overloading the input of your DAC when boosting any frequencies… When cutting frequencies, you will need to compensate in the other direction to maintain a balanced output level… So when boosting some frequency range(s) and cutting some frequency range(s) you must monitor your output level of the EQ plug-in to maintain an equilibrium where the output signal level is always near 0dB or -3dB to -6dB less… It is called “Gain Management” … Ultimately you don’t want to add more gain to the digital-audio data signal that is being input to and output from Audirvāna… This also needs to consider the upsampling gain in aggregate.
Great explanation. I get it. Thanks.