Audio Asylum Thread Printer

Get a view of an entire thread on one page

FAQ / News / Events

Digital Drive

RE: Your claim is completely incompatible with his statement

77.59.161.90

In Reply to: RE: Your claim is completely incompatible with his statement posted by Chris from Lafayette on June 19, 2024 at 11:34:34

Read the papers and slide presentation and answer the questions for yourself.

He is talking about an individual response. Taken in aggregate you will build a statistical preference for most listeners (assuming you have a good correlation to the measured data) for particular types of distortion vs. other types.

Geddes papers (if you don't read them then we have nothing more to discuss) indicate that THD and IMD had no correlation (even a slightly negative slope) to the distorted sounds they were subjecting listeners to. When he applied his metric, which takes into account some kind of weighting for harmonic content, he got a much stronger and positive correlation between preference and his metric.

Follow Ups:

I think GedLee did his best work in Rush - 13th Duke of Wymbourne 12:02:45 06/20/24 (1)
I scanned through the two AES papers and the conclusion is convincing that THD and IMD doesn't match the subjective scoring of their test group whereas their Gm metric (bad choice of symbol as Gm means something else to a lot of people) does correlate much better. There is much to pick over - in the first part they don't quantify masking. Yes it is a psychoacoustic thing but how much effect does it have and why choose a cos^2 weighting factor? I also question one of their precepts that distortion will be more apparent at low signal levels than high signal levels - that needs some explaining as distortion grows quicker than the signal level that causes it (i.e. if a signal increases by 1dB any second harmonic will grow by 2db, third harmonic by 3dB etc).
We aren't told what the 21 transfer functions are nor their correspondance to trial number. We do know one/some/many have severe discontinuities - are those really relevant for high-end audio? And they remove 6 of those transfer functions at the end because they don't fit the scatter plot so well (at least they were open enough to admit this as data manipulation is often a big problem in academic papers).
But, the big unanswered question is whether a low distortion system is somehow worse? We don't know because they don't detail the transfer functions! We can probably conclude that many non-linearities are not that audible, or audible at all, but we can't conclude that low distortion devices are somehow subjectively inferior. So it is a bit of a shield for audiophiles to stand behind and an excuse for high-end audio designers to not work too hard but not the full story.

Edits: 06/20/24
- RE: I think GedLee did his best work in Rush - morricab 05:45:55 06/21/24 (0)
  Indeed! Rush rules!
  
  Cheever added an SPL factor into his model because there is also a change in the ear's own generated harmonics. It also then would depend on the sensitivity of the speaker and how much distortion/power at a given SPL...
  
  So, a high sensitivity speaker that only needs mW to a couple of watts will be an easy, low harmonic dominated (which are basically masked) distortion pattern that will sound clean.
  
  Once you get beyond a certain SPL it will really depend on how rapidly the devices distortion, particularly the high order harmonic distorion rises and how that still compares to the ear/brain masking...complicated but can go a long way to explaining a lot of the subjective varaiabilty observed.
  
  "But, the big unanswered question is whether a low distortion system is somehow worse? We don't know because they don't detail the transfer functions! "
  
  It is not low distortion per se, it is the fact that there are high order harmonics without lower order harmonics to provide masking. The audibility seems to be down to very low levels. Achieving this with negative feedback has it's issues with regard to artificial noise floor (read Crowhurst and Pass white paper) and IMD that gets generated. Just seeing absent harmonics in a static test does at all tell the whole story.

Post a Followup:

FAQ

Post a Message!

Forgot Password?

Moniker (Username):
Password (Optional):
	Remember my Moniker & Password (What's this?) Eat Me
E-Mail (Optional):
Subject:

Message:	(Posts are subject to Content Rules)

Optional Link URL:
Optional Link Title:
Optional Image URL:
Upload Image:
E-mail Replies:	Automagically notify you when someone responds.

To view your new posting or follow-up, click on the RELOAD or REFRESH button on your browser.

This post is made possible by the generous support of people like you and our sponsors: