Possible to rank equipment based on what we actually hear and perceive?

[richard12511]

Perhaps the question could be rephrased: at what point does noise and distortion fall below the threshold of audibility?

The entire point of electronic measurement is to set a standard.https://www.nist.gov/ctl/pscr/speech-intelligibility-demo

I'm curious what my previous Yamaha entry AVR was, as I wasn't able to hear any difference between it and a near SOTA Nord Hypex amp under blind conditions. Same thing with the other listeners.

If I had to guess, I'd say anything beyond -70dB or so is equivalent to my ears. If you can get to -70dB, you get 10/10 for distortion performance. After that I'd be looking to compare max output, ability to handle tough loads, frequency response(usually 10/10), fan noise, and price.

 

[MaxBuck]

My subjective ranking is that my equipment is best. The rest of you can only hope to match it.

 

[Chromatischism]

I'm curious what my previous Yamaha entry AVR was, as I wasn't able to hear any difference between it and a near SOTA Nord Hypex amp under blind conditions. Same thing with the other listeners.

What speaker was being used?

 

[richard12511]

What speaker was being used?

Infinity Beta 20 I believe.

 

[Chromatischism]

Infinity Beta 20 I believe.

Hmm they claim 8 ohms but I really doubt that.

 

[Jim Matthews]

I'm curious what my previous Yamaha entry AVR was, as I wasn't able to hear any difference between it and a near SOTA Nord Hypex amp under blind conditions. Same thing with the other listeners.

If I had to guess, I'd say anything beyond -70dB or so is equivalent to my ears. If you can get to -70dB, you get 10/10 for distortion performance. After that I'd be looking to compare max output, ability to handle tough loads, frequency response(usually 10/10), fan noise, and price.

I suspect that further waveform study (as started by Serge Smirnoff "Diffrogram") will bear on what we can't define as a good device, if the results can be reproduced and applied.

While it's clear that a device that tests better in SINAD ranking has a measurable improvement over lesser ranked gizmos it may just be a proxy for what we really notice in Transient distortions, particularly in the time domain.

I have always been suspicious of a new measurement that dismisses what was perfectly satisfying (like your Yammie AVR) due to escalating performance benchmarks.

 

[preload]

"Possible to rank equipment based on what we actually hear and perceive?"

Yes and for DACs, amps, and other SOTA devices the overwhelming majority are going to have identical "ranks." If you haven't already arrived at that conclusion yet, I'm saving you a whole bunch of time. You're welcome.

 

[Sal1950]

Some preamps get quieter the further you turn the volume, others, it is the opposite.

John, Can you explain that a bit?
I would think the noise would track volume/gain level of the device.
Noise at minimum at minimum volume, vise versa?

 

[rdenney]

John, Can you explain that a bit?
I would think the noise would track volume/gain level of the device.
Noise at minimum at minimum volume, vise versa?

Johnson noise in the potentiometer that is most volume controls. At zero ohms, the noise is low (not that it matters), and at infinite ohms, the noise is low (not that it matters). It's in between that it can vary, and it also depends on the circuit design as to which way the noise trends with the volume settings. But at low volume, typical preamp pots are at high resistance, and can produce greater thermal noise (in absolute terms) as a result.

Rick "received knowledge" Denney

 

[dwkdnvr]

When I saw the thread title, I thought it might be going in a different direction. I've wondered whether it was possible to evaluate the audibility of the 'error signal' of a component by leveraging the work done by the perceptual coding guys. The basic idea is that if a component is audibly transparent, then if you 'perceptually encode' the measured output of the component it should be 'the same' as the encoding of the original track. Conversely, if there are audible deviations introduced by the component, then the perceptual encoder would have to include those deviations and you should have a larger encoded result.
To be really useful you wouldn't just throw the output into say an MP3 encoder but instead I'd imagine you'd crack open the algorithm to expose various internal metrics and filters.

I'm not up on my perceptual coding enough to know whether this is actually in any way viable or is just a dumb idea, but it sounds reasonable when I run it through my head