Benchmark Media blog post about distortion

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Yes, the point I was making is that the idea that distortion which is below the threshold of audibility is necessarily inaudible relies on there being no 'anti-masking'. That is, a sound below the threshold of audibility cannot be rendered audible by the presence of some other sound. Naively one might expect this to be true (given that masking has the opposite effect), but who knows whether there is some sort of parametric amplification possible in the ear or brain with certain combinations of frequencies?


Thanks. I was not aware of that, but it makes sense. Just to clarify, this is a sound which would be audible if not masked, but it has been masked by a second sound, but is then rendered audible again by a third sound?

Well the whole concept of "Threshold of audibility" needs to be examined, I believe.
A couple of points to ponder:
- remember loudness & hence audibility threshold is a perception, not an engineering measurement!!
- it always implies that there is some masking taking place as no audibility test is without noise - so we are always looking at what is audible in the current noise environment.
- The thresholds often referred to are the Fletcher-Munson equal loudness curves but I believe these were taken a long time ago & with mono single tones. There have been updates since then which changed these curves but I'm more interested in what is audible in binaural listening to complex signals - this is much more like music listening.
- AFAIK, we sense the total power in each ERB as the loudness according to JJ Johnston
  • loudness is approximately (sometimes very approximately for some special signals) proportional to the 1/3.5 power of the signal power, or the 1/1.75 power of amplitude."
  • Signals in the same “band” convert intensity to loudness approximately with the power law relationship given in the last slide.
  • Signals in different bands ADD linearly in loudness
.

One example of this is that the equal loudness curves for broadband noise are about 10dB lower than the F-M curves - in other words we hear noise at lower thresholds (10dB lower) than we do individual tones. I suggest this is because of the total power in the ERB is being sensed as loudness

So in relation to your question; could two or three or more sounds which fall into the same ERB frequency band but are inaudible as individual tones, become audible when combined? It seems reasonable to assume this but I don't know of research which shows this.
 
Last edited:
AX tech editor
Joined 2002
Paid Member
Even a young and undamaged ear has much more deeps and peaks than +/- 1dB,[snip]
Can we measure distortion caused by of our ears?
I haven't find information about it yet.

Yes, but when you listen to live sound you also have the ears in between. Your idea of live sound includes the response of your ears.
So when trying to reproduce as good as possible it is sensible to strive for low distortion and flat response, even if your ears are not, so that it sounds as close as possible to the original.

Jan
 
scottjoplin said:
Lots of mights, coulds and maybes here, interesting, but why would one want to make inaudible distortions audible or is it purely academic?
The point is that the 'many zeros' brigade may justify their quest for very low distortion on the grounds that distortion below the audible threshold is by definition inaudible, so their 'many zeros' equipment is audibly perfect. The existence of 'anti-masking' blows a hole in this. It could encourage them to seek for even more zeros.

On the other hand, most evidence is that few or no zeros are needed.
 
Would be interesting to understand the evidence. References, please (pretty)?

//
Some reading on equal loudness curves fro tones Vs noise Equal-loudness contour
"he BBC research department conducted listening trials in an attempt to find the best weighting curve and rectifier combination for use when measuring noise in broadcast equipment, examining the various new weighting curves in the context of noise rather than tones, confirming that they were much more valid than A-weighting when attempting to measure the subjective loudness of noise. This work also investigated the response of human hearing to tone-bursts, clicks, pink noise and a variety of other sounds which, because of their brief impulsive nature, do not give the ear and brain sufficient time to respond. The results were reported in BBC Research Report EL-17 1968/8 entitled The Assessment of Noise in Audio Frequency Circuits."

This is the standard that came out of this BBC research ITU-R 468 noise weighting - Wikipedia
400px-Lindos3.svg.png


"The CCIR curve differs greatly from A-weighting in the 5 to 8 kHz region where it peaks to +12.2 dB at 6.3 kHz, the region in which we appear to be extremely sensitive to noise."

Edit: "468-weighting is also used in weighted distortion measurement at 1 kHz. Weighting the distortion residue after removal of the fundamental emphasises high-order harmonics, but only up to 10 kHz or so where the ears response falls off. This results in a single measurement (sometimes called distortion residue measurement) which has been claimed to correspond well with subjective effect even for power amplifiers where crossover distortion is known to be far more audible than normal THD (Total harmonic distortion) measurements would suggest."
 
Last edited:
Sorry I don't have a reference to hand, but my understanding is that tests about 60 years ago showed that most listeners cannot hear anything less than about 0.5% low order distortion on music. It may be that more recent tests have pushed the level down a bit, although tests on 'live vs. reproduced' seem to be quite thin on the ground. Most modern tests seem to be 'this vs. that' where both are reproductions.
 
Yes, the point I was making is that the idea that distortion which is below the threshold of audibility is necessarily inaudible relies on there being no 'anti-masking'. That is, a sound below the threshold of audibility cannot be rendered audible by the presence of some other sound. Naively one might expect this to be true (given that masking has the opposite effect), but who knows whether there is some sort of parametric amplification possible in the ear or brain with certain combinations of frequencies?


Thanks. I was not aware of that, but it makes sense. Just to clarify, this is a sound which would be audible if not masked, but it has been masked by a second sound, but is then rendered audible again by a third sound?
Research which directly answers your points
"the hearing of a complex tone is influenced by a higher harmonic component though the component is not audible."
From "Study on Perception of Complex Low Frequency Tones"

6. CONCLUSIONS
The hearing thresholds and the perception frequencies of complex frequency tones
with higher harmonics in frequencies between 5Hz and 50Hz were measured. The
following conclusions were obtained.
(1) Hearing thresholds of the complex tone with harmonics decrease compared to
the hearing threshold of the pure tone, whose frequency is the same as the
fundamental.
(2) As the number of harmonics increases, the hearing thresholds of complex
tones decrease in many cases.
(3) A harmonic below the hearing threshold affects the hearing of a low
frequency complex tone.
(4) The hearing threshold of a low frequency complex tone is affected by the
phase difference between components
 
Last edited:
The point is that the 'many zeros' brigade may justify their quest for very low distortion on the grounds that distortion below the audible threshold is by definition inaudible, so their 'many zeros' equipment is audibly perfect. The existence of 'anti-masking' blows a hole in this. It could encourage them to seek for even more zeros.

On the other hand, most evidence is that few or no zeros are needed.
Some evidence that many zeroes are needed when differences at -110dB -120dB seem to be perceptible
AKSA's Lender Preamp with 40Vpp Output

Edit: As I said before one has to be careful about the received wisdom of audibility thresholds as are often used to dismiss differences in measurements
 
Last edited:
mmerrill99 said:
Some evidence that many zeroes are needed when differences at -110dB -120dB seem to be perceptible
Actually that is more like -80db -90dB because the fundamental is around -20dB. Two zeros after the decimal point. Of course there is only anecdotal evidence of the change being audible. Interesting that the preferred version has higher distortion, but with 2nd dominating.
 
Actually that is more like -80db -90dB because the fundamental is around -20dB. Two zeros after the decimal point. Of course there is only anecdotal evidence of the change being audible. Interesting that the preferred version has higher distortion, but with 2nd dominating.
Yes, you are correct - fundamental @ -20dB.
Yes, 2nd has doubled from 0.0021% to 0.0046% but 3rd has decreased from 0.0031% to 0.0009%

Sure, it's anecdotal - choose your poison!!
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.