I spent some time playing around with ideas mentioned in previous posts about the -1 dB/octave rolloff presumed for an ideal listening room. Screenshot of the setup here…
The processing is hosted by the Juce Plug-In Host (lower right) provided in the JUCE Examples folder. This is a neat little plugin environment that gives you control over AU, VST, and VST-3 plugins, as well as allowing easy setup of the audio I/O devices. I highly recommend this little tool. (It’s free!)
My processing chain runs from audio input through a pair of iZotope Ozone 7 Equalizers, then into vTuning, and then through another EQ to the audio output. I have my system set up to assume its default Audio Out is SoundFlower virtual audio cable. The input to the plugin host is the other end of the SoundFlower port, and its audio output is being sent to my MOTU Ultralight USB Audio Interface for my headphone listening. This arrangement allows me to feed all audio from Web browsers, iTunes, and Spotify through my corrective system running in the background on the computer.
The EQ panels are shown on the left side of this screenshot. The top EQ provides a roughly -1 dB/octave rolloff on the incoming audio, with the presumption that such is the intended listening environment. And my headphones might be capable of providing that kind of rolloff on their own. But the trouble for me is that my hearing impairment won’t allow me to hear that rolloff the same way a person with unimpaired hearing would hear it.
So we provide the rolloff of the ideal listening room up front, ahead of Crescendo processing so that I can be helped to hear it as it should be heard.
But, in turn, Crescendo assumes it is playing into a spectrally flat output transducer. And so in order to flatten my headphones I provide the opposite rolloff with a +1 dB/octave EQ on the bottom left.
The EQ panel in the middle is for my personal hyper-recruitment and decruitment hearing treatment – again, applied ahead of Crescendo processing. My hyper-recruitment needs a mild dip of 3 dB around 1800 Hz, and a mild 4 dB shelf above 3 kHz. This works incredibly well, but it is a topic for another post. So for this discussion, just consider that hyper-recruitment EQ and the vTuning plugin as constituting my Crescendo processing. We want to focus on the -1 dB/octave rolloff going into the Crescendo processing, and the headphone flattening EQ coming out of the processing.
The EQ for headphone flattening isn’t a precision device, just an approximation opposite of what the headphones do to the sound. But it seems close enough.
While listening to some tracks I must say that the results are quite good sounding. Are they correct? What is correct? I do hear interesting details that cannot be heard directly by me, without this processing. It doesn’t sound overly bright, but yet there are high frequency details – cymbals, higher order harmonics, more accurate instrumental timbres. Without my processing the music would have a dull, dark, basement quality, since my hearing is nearly a brick wall filter cutting away everything above C5, an octave above middle-C.
Every equalizer plugin has its own way of specifying its filter parameters. For the iZotope 7 EQ, in order to achieve the approximate 1 dB/octave rolloffs, I used a low shelf at 183 Hz with the lowest possible Q (here 1.0), and a shelf boost of 1.5 dB. The converse high shelf is at 892 Hz with 1.5 dB dip. That seems to give a total rolloff of about 6 dB from 40 Hz to 10 kHz. What happens above 10 kHz is of no consequence for me. But that corresponds to about 7 octaves, so we have pretty close to 1 dB/octave rolloff. Feel free to experiment with variations – nothing is cast in stone.
[ Note: I also have the input gain on the bottom EQ set to -18 dB, to give Crescendo the necessary headroom to avoid clipping. I make up for that attenuation with the volume control knob on my headphone amplifier, so that at an RMS level of -16 dBFS, I’m getting about 73 dBSPL out of the headphones, with a 1 kHz test tone.]
And while my two EQ’s (top and bottom) seem like mirror images of each other, they really have quite different purposes. Setting them up as a mirror image only worked for me because my Sennheiser HD650 headphones produce something reasonably close to 1 dB/octave rolloff on their own. But recall that the top EQ is intended to establish the ideal listening room characteristic rolloff. The bottom EQ is intended to cancel out any rolloff already present in the headphones so that Crescendo has a spectrally flat transducer to play into.