On a 2010 Mac Mini OS 10.6.8, Audacity 2.2.2. Listening on Sony studio monitor headphones.
It appears there’s a limit on dynamic range loudness when EQ’ing these brickwalled files even in 32 bit floating point. Reducing amplification by 3db doesn’t help. Also this playback distortion doesn’t happen all the time depending on the song and the level of brickwalling and/or whatever software was used by the mastering engineer to make the song louder usually at an average of around -12dbRMS according to Wave Stats.
I use Apple Core Audio based Audio Units Graphic EQ plugin. Works very good most of the time. Even though I hear the distortion adjusting EQ sliders, when I apply and of course cause clipping and then apply Amplify to correct, then apply my own Limiter to regain the loudness, there is no distortion playing back in Audacity. It only happens editing with the AU EQ plugin.
Also this got me to question what I should set my MIDI headphone output to for bit level and sampling rate. I just learned in the past weeks this Apple Audio Device Utility is adjustable and so thought the DAC on my sound card and its MIDI setting is the source of the distortion. Any thoughts on this?
I have Audacity Real Time Conversion set to best quality, Dither: None. Changing it doesn’t seem to make a difference.
If I’m just pushing the limits of my DAC is there a way to calculate how far to go when applying an EQ in Audacity?
Thanks for the link, but that doesn’t explain why I get distortion during live edit playback, where the distortion isn’t in the final limited and saved to hard drive finished AIFF file.
You’re on Windows so do you have similar bit/sampling rate MIDI adjustments for headphone output as I described above on the Mac?
32 bit floating point won’t clip “internally” and a floating-point WAV or AIFF won’t clip, but you can clip your DAC during playback. Or, you can clip your analog amplifier. And, if you have Audacity configured to show clipping, it will show potential clipping if you go over 0dB.
It appears there’s a limit on dynamic range loudness when EQ’ing these brickwalled files even in 32 bit floating point. Reducing amplification by 3db doesn’t help.
How much did you boost with EQ? That’s not always a 1:1 effect… Actually, cutting with EQ can sometimes boost some peaks because of phase shifting and the fact that filters are imperfect
or whatever software was used by the mastering engineer to make the song louder usually at an average of around -12dbRMS according to Wave Stats.
Clipping is related to the peaks, not the RMS.
It only happens editing with the AU EQ plugin
It’s possible that plug-in can’t go over 0dB, but I’d say that’s unlikely.
If I’m just pushing the limits of my DAC is there a way to calculate how far to go when applying an EQ in Audacity?
Your DAC will clip at exactly 0dB. 0dB is defined as the maximum you “count to” with a given number of bits. Your drivers scale the bit-depth to match your hardware so a 0dB 8-bit file plays back at the same volume as a 0dB 24-bit file, even though the 24-bit file has bigger numbers.
Floating-point data uses a different standard. With floating point a value of 1.0 is 0dB. So the numbers are much-much smaller but again your drivers scale-up when converting to integer to match your hardware.
According to the Amplify effect around 8 to 12 db which required two EQ/Amplify effect go rounds due to the distortion in live playback editing with the second reducing the waveform to -17db on some songs which of course gave me back some headroom where I could just increase the Mac’s system volume slider so I could hear what I was editing with the second EQ. I tend to forget my ears adapt to loudness levels during long EQ edit sessions and I think I just went too far. These are all subjective based edits that often wind up sounding the same as the original unedited version after the limiter is applied.
But your post made a lot of sense in indicating I was pretty much pushing the DAC too far. But I’ld still like to know what to set the MIDI bit/sampling rate for headphone output or at least confirm whether this is necessary. When I first bought and setup my 2010 Mac Mini in 2011 I didn’t even know about the MIDI or that it had anything to do with live playback working in a DAW. I’ve left the headphone output at 24 bit/44Khz (default) in the MIDI setup interface. I can increase to 32bit/96KHz. Should I do this?
So to clarify I think I can solve the live playback distortion by adjusting the listening volume that doesn’t get me to over drive the EQ which some sliders I’ve gone above 6db in the midrange/highlights area with the kick drum 50Hz going over 9db. I can stop the distortion and still get what I want with EQ by doing several EQ/Amplify reduction/add another EQ edit/adjust OS volume slider iterations.
I wish there were lessons or tips on how to start out editing CD rez audio to make it sound bigger than life. A lot of the CD pop music recorded in the '70’s and even the '80’s including several remastering versions by the label still winds up sounding thin with a weak or nonexistent kick drum bottom end.
Apple’s Core Audio (hardware/OS level integration) install of Audio Units Graphic EQ (32 bands) is a real-time equalizer. It appears from what I’ve gathered from online discussions of Q-factor in relation to octave bandwidth EQ affect with each band slider is around a Q of 1 or 2 but it’s kinda’ hard to tell. This Apple EQ plugin Q-factor is fixed and non-adjustable.
I think I might have found the problem and the solution.
Just by sheer luck I happened upon an Apple AU GraphicEQ YouTube tutorial where the instructor preferred a 10 band setting. So I switched to 10 bands instead of 31 and the audio just became much more responsive to what I was wanting especially in the bottom end kick drum. Before it was like wack-a-mole tapping down too much bass resonance using 31 bands.
It only required one EQ adjust right after the AUMatrixReverb plugin and I was able to open up the sound and make it big without distortion. All tweaks were so much smoother and less jumpy.
I think hacking away with 31 bands was taxing the processor into not being able keep up. What a world of difference in the sound I got with just 10 bands.