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.
There are various things that could be said about that article.
Please do! That's why I started the thread! Well, that, and the fact that I've never started a thread here before, and so figured it was about time. :p

I would just note that not all pianos are stretch-tuned, although it does make them sound better (IMHO).

I know more than one piano tuner who would go so far as to say that a piano that hasn't been stretch-tuned, hasn't been tuned.
 
I think all he was saying is that if distortion is below the theshold of hearing than it is by definition inaudible even without any masking. This may or may not be true. If it is true then sufficiently low distortion is inaudible, so the equipment would be audibly perfect (as far as nonlinear distortion is concerned). This can then be used to justify lots of 0's after the decimal point.

Two snags with this approach:
1. it assumes that there cannot be any 'anti-masking' i.e. a sound makes other sounds easier to hear - this may be true but I am not sure it has been exhaustively demonstrated
2. optimising one aspect of performance (in this case, nonlinear distortion) often means compromising on something else either because physics demands it or because the designer's attention was diverted
 
I think all he was saying is that if distortion is below the theshold of hearing than it is by definition inaudible even without any masking. This may or may not be true. If it is true then sufficiently low distortion is inaudible, so the equipment would be audibly perfect (as far as nonlinear distortion is concerned). This can then be used to justify lots of 0's after the decimal point.

Two snags with this approach:
1. it assumes that there cannot be any 'anti-masking' i.e. a sound makes other sounds easier to hear - this may be true but I am not sure it has been exhaustively demonstrated
2. optimising one aspect of performance (in this case, nonlinear distortion) often means compromising on something else either because physics demands it or because the designer's attention was diverted

Yes, the part I bolded is true - it's called comodulated masking release which does exactly what it says on the tin - when another frequency synchronously modulates with a masked sound it un-masks the masked sound
 
It seems like two different things people are talking about.

"Comodulated masking release" is an effect that reduces a masking effect in the masker is modulated. Comodulation Masking Release In Electric Hearing

However, DF96 suggested anti-masking would be an effect that would make a sound that was below the baseline threshold of hearing when unmasked, more audible in the presence of, or when summed with, an anti-masking sound. The idea there might be that a sound below the threshold of hearing might be inaudible by itself, but maybe adding it to another sound could result in a composite (summed) sound that makes the low level sound more audible. If one thinks about the waveforms both for individual sounds and the summed sound and the instantaneous SPL for each over time, for the summed signals the instantaneous SPL of the low level signal could at times be quite high (even though it's variation might be very small), perhaps at an absolute SPL that would make it more audible. If the anti-masker sound was chosen to minimize normal masking of the low level sound maybe it could work, I don't know. Also, I don't know if I explained the idea as I understand it very well. DF96, were you thinking of something like what I just described?
 
It seems like two different things people are talking about.

"Comodulated masking release" is an effect that reduces a masking effect in the masker is modulated. Comodulation Masking Release In Electric Hearing[

However, DF96 suggested anti-masking would be an effect that would make a sound that was below the baseline threshold of hearing when unmasked, more audible in the presence of, or when summed with, an anti-masking sound. The idea there might be that a sound below the threshold of hearing might be inaudible by itself, but maybe adding it to another sound could result in a composite (summed) sound that makes the low level sound more audible. If one thinks about the waveforms both for individual sounds and the summed sound and the instantaneous SPL for each over time, for the summed signals the instantaneous SPL of the low level signal could at times be quite high (even though it's variation might be very small), perhaps at an absolute SPL that would make it more audible. If the anti-masker sound was chosen to minimize normal masking of the low level sound maybe it could work, I don't know. Also, I don't know if I explained the idea as I understand it very well. DF96, were you thinking of something like what I just described?

There's a number of things to address in your reply but first I read DF96's text again & don't get the meaning from it that you do
"1. it assumes that there cannot be any 'anti-masking' i.e. a sound makes other sounds easier to hear - this may be true but I am not sure it has been exhaustively demonstrated"

From this I understood he was saying "is there an added sound which can unmask a sound "a sound makes other sounds easier to hear"? I can't see this referring to sounds below the ABSOLUTE threshold of audibility rather it concerns sounds which are being masked by noise or some other signal & therefore inaudible.

Sorry but the paper you linked does not state what you claim ""Comodulated masking release" is an effect that reduces a masking effect in the masker is modulated"

From the paper you linked "Comodulation masking release (CMR) is an improvement in the detection threshold of a masked signal that occurs when the masker envelopes are correlated across frequency (i.e., comodulation). CMR can be observed when flanking bands (FBs) of noise co-modulated with an on-frequency band (OFB) noise masker are added at remote frequencies (CMR1)"
Which is exactly what I stated! It's not about modulating the masker signal (CMR) but rather a signal at a different frequency (CMR1)

But your statement is also another form of CMR - where a decrease in masked thresholds occurs when the masker is amplitude-modulated. Comodulation Masking Release | Auditory Neuroscience

I'm not sure what you think he said but there is a related point - There is a whole area to be discussed about the relationship between dB, SPL & loudness perception. Perception of loudness is related to energy in the critical bands or ERB of the hearing mechanism. The consequence of this is that equal loudness curves from Flecther Munson are higher than equal loudness curves for noise.

In other words thresholds for noise are lower (by about 10dB, I think) than the F-M curves.

Why this is? Because the total energy in the ERB is higher for noise (mixed frequencies) than it is for pure tone that falls in the same ERB.

So can a number of inaudible tones that all fall into the same ERB become audible due to the total energy in that ERB becoming discernible?
 
Last edited:
AX tech editor
Joined 2002
Paid Member
1. it assumes that there cannot be any 'anti-masking' i.e. a sound makes other sounds easier to hear - this may be true but I am not sure it has been exhaustively demonstrated

It has been demonstrated that in some cases, where a tone is not audible, it can be 'unmasked' if another tone is added.
But I don't have the ref ready right now, I'll see if I can find it tomorrow.

Edit: It's complicated ;-)

Edit edit: Mark beat me to it...

Jan

ABSTRACT
A forward‐masking situation was investigated where more energy added to the masker decreased the threshold of the signal. The signal was a single sinusoid. The masker contained two frequency components: one at the same frequency as the signal, and one at a variable frequency. For certain frequencies and intensities of this variable masker component the threshold of the signal was lower than if that component were not present.This decrease in threshold is unmasking. Unmasking is interpreted as being caused by the suppression of the fixed masker component by the variable masker component, thus producing less masking at the signal frequency. The basic pattern of masking and ummasking was qualitatively similar to the pattern of pulsation thresholds observed by Houtgast [Acustica 29, 168‐179 (1973). Unmasking was not observed at absolute threshold or in a simultaneous masking situation. Unmasking was observed at all signal frequencies tested (0.5–6.0 kHz) and was not dependent on the phase of the masker components or the phase between the masker and the signal. When the frequency of the variable masker component was less than the signal frequency, unmasking was dependent on only the intensity of the variable component. However, when the variable frequency was greater than the signal frequency, unmasking was dependent on the difference in intensity between the two masker components. The existence of unmasking in a forward‐masking situation does not depend on whether or not the masker and signal are turned on and off abruptly or smoothly. These characteristics of unmasking are not in conflict with the assumption that this effect is caused by the variable masker component suppressing the other masker component.
 
Last edited:
Here you go: Comodulation Masking Release | Auditory Neuroscience
"Comodulation Masking Release (CMR) is the decrease in masked thresholds that occurs when the masker is amplitude-modulated."
As I said, it also occurs when sounds at different frequencies co-modulated with the masking frequency. These frequencies wouldn't be called maskers in the normal sense as they are remote in frequency from the signal being detected to act as maskers so that definition from Auditory neuroscience is better explained by the diagram on that page where it shows that a modulated masker becomes more effective at releasing masking as the bandwidth of the masker is increased.

AuditoryNeurosciFig6-1.JPG


As you can see from the diagram, the threshold for detection is the same up to about 100Hz bandwidth of masker - whether it is modulating or not modulating doesn't matter, threshold is unchanged - it's the increasing bandwidth that makes the difference in other words it's the remote frequency modulating signal that makes a differences as I said originally
 
Just to explain further - the "CO" of the term co-modulation masking release is the important word fragment. It means that something else is modulating in synchrony with the masking tone NOT that th emasking tone itself is modulating.

When the other sound that is modulating is in a different critical band or ERB to the tone being masked, it reduces the amplitude threshold that is detectabe.
 
Markw4 said:
However, DF96 suggested anti-masking would be an effect that would make a sound that was below the baseline threshold of hearing when unmasked, more audible in the presence of, or when summed with, an anti-masking sound. The idea there might be that a sound below the threshold of hearing might be inaudible by itself, but maybe adding it to another sound could result in a composite (summed) sound that makes the low level sound more audible.
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?

mmerrill99 said:
Yes, the part I bolded is true - it's called comodulated masking release which does exactly what it says on the tin - when another frequency synchronously modulates with a masked sound it un-masks the masked sound
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?
 
I'm wondering about it for a long time, we are worrying about frequency response of our audiosystem. We say for example, what a crap this power amplifier, it has a frequency response only +/- 1dB between 20Hz-20kHz.
Loudspeakers has a much more more deviances in this regard.
And how's about our ears?
I think this is the more important part of our audio system.
Even a young and undamaged ear has much more deeps and peaks than +/- 1dB, just for example as for my daughter as you can see in attachment:

http://www.diyaudio.com/forums/attachments/the-lounge/573558d1475948702-actual-hearing-examination-results-hearing-panni-jpg

And how's about distortion?
When my right ear was damaged lately because of unknown reasons, one of my complains was about higher perceived distortion.
Can we measure distortion caused by of our ears?
I haven't find information about it yet.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.