Human Sensitivity to Frequency Response

[ahofer]

I've searched this and not found a thread elsewhere. In another forum, someone posted this:

(https://www.harbeth.co.uk/usergroup...stimulate-discussion.80476/page-2#post-101241)

And they go on to interpret it as:

the curves appear to suggest that the JND [Just Noticeable Difference] for a deviation from flat (peak or dip not specified) with a centre frequency of, say, 200 Hz would be about 2.7dB where the deviation extends over 1/3 octave, about 1.2dB where it extends over an octave and about 0.2dB where it extends over 3 octaves.

I hadn't seen this before, and it certainly calls attention to frequency response issues in the presence region and above. In particular, I understand why some speakers with 3db of emphasis broadly above 7Khz (I'm calling it showroom treble) are so irritating.

I suppose the graph feels a little misleading, as a three octave Q around 5K spans 2.5kHz to 10kHz, whereas a 3 octave Q around 50 is 25-100, a range in which sensitivity varies significantly. I presume the sensitivities are calibrated to the center of the dip/emphasis?

This is prompted by Alan Shaw's "Tech Talks", which are pretty good for vendor-sponsored education.

 

[Blumlein 88]

I don't think it is misleading. It matches my experience. Very slight EQ changes over octaves makes for a noticeable difference in sound that many people don't expect. I've blind tested myself on microphone preamps, and DACs that have .3 or .4 db droops over the 3 octaves at both the top and bottom of the scale and they are discernable.

For instance some devices will say they are -3db down at 20 hz. Such a device isn't within .1 db of flat until 200 hz. (Think old analog tape decks) That level vs one which is -3 db down at 2 hz is something you can notice. Same at the other end of the spectrum. A little droop in the top couple octaves does sound a bit soft in the treble.