Re: Nice dodge
And you are being your usual disagreeable self.
I did not see a "question re: audibility", only statements about nonlinearities and "Ignoring the actual audibility of the products involved", which, to me, is the whole point, not something to ignore.
Having designed numerous microphones in my time, I am well aware of the nonlinearities. They are always very low order, the backplate nonlinearity being domninately second order. This would make the distortion virtually inaudible. The electronics for a microphone are usually biased class A types with low crossover distortion, but often a soft clipping function resulting from the FET limitations. This too is of a very benign nature. From my perspective there might be high THD or IMD NUMBERS, but there is no data to say that any of this would be audible. Indeed, all indications and research points in the other direction - that it is not audible.
auplater said:
And you are being your usual disagreeable self.
I did not see a "question re: audibility", only statements about nonlinearities and "Ignoring the actual audibility of the products involved", which, to me, is the whole point, not something to ignore.
Having designed numerous microphones in my time, I am well aware of the nonlinearities. They are always very low order, the backplate nonlinearity being domninately second order. This would make the distortion virtually inaudible. The electronics for a microphone are usually biased class A types with low crossover distortion, but often a soft clipping function resulting from the FET limitations. This too is of a very benign nature. From my perspective there might be high THD or IMD NUMBERS, but there is no data to say that any of this would be audible. Indeed, all indications and research points in the other direction - that it is not audible.
The question of audibility is totally fair and extremely significant.
However, when I said 'ignoring the audibility of the products' I was specifically refering to the products themselves under measurement (ie. sum/difference frequency components).
If we make a difference-frequency analysis of a compressor in 'typical use', we will see the same products as those made in the mic capsule, whose level we can vary with the compressor settings.
However, while we can easily argue & test whether such products are audible at various constant listening levels (of sinewaves), how we perceive 'compression' itself, applied to music, is a different matter (a complicated one!).
As to the audibility of this type of non-linear distortion, there is a well known paper, often cited, that showed the audibility of far smaller levels of distortion:
- from the AES online abstract search -
"
Threshold of Audibility of Transient Intermodulation Distortion
The audible threshold of TIM was determined for six most sensitive subjects of the previously reported test series of 68 listeners. Improved equipment, carefully controlled listening environment, a digital TIM generator, and five stereophonic music samples were used. The results show that in certain passages of music, 0.003% of distortion is clearly audible....
Paper Number: 1392
AES Convention: 61 (November 1978)
Authors: Petri-Larmi, Margit; Otala, Matti; Lammasniemi, Jorma
"
Clearly, this work was aimed at power amplifiers but is no less applicable to microphones.
While we use constant sinewaves in the difference frequency test, we are only looking for an indication of the significance of the non-linearities involved at a given SPL.
How these non-linearities will be perceived in the case of music - the context of transient behaviour - is an entirely different matter, and to this I would suggest looking at the difference frequency distortion product thresholds shown at levels of compression which are audible as 'compression' for guidance.
As to the empirical audibility of the distortion present in the average studio condenser mic, when compared with one of my acoustic impedance matching mics on high-SPL sources such as a drumkit or orchestra the difference is extremely audible.
Andy
audibility
Thanks for the clarification Andy, and resolving your own inquiry. It's refreshing to hear an informed alternative to the usual drone.
I'm especially interested in what's occurring during transient high level attacks on , for instance, staccato piano notes, drum kit, and orchestral peaks in complex musical passages. What I often hear on specific tracks across many reproduction platforms transcends the particular equipment and seems to reside in the recording itself, which may implicate the mic-ing technique, quality, recording chain, etc.
Piano solos seem especially vulnerable to a, for lack of a better description, "gritty" "pumped" sound that is exceedingly annoying. I would guess this may very well be compression induced distortion occurring in the microphone, and is highly audible.
John L.
Thanks for the clarification Andy, and resolving your own inquiry. It's refreshing to hear an informed alternative to the usual drone.
I'm especially interested in what's occurring during transient high level attacks on , for instance, staccato piano notes, drum kit, and orchestral peaks in complex musical passages. What I often hear on specific tracks across many reproduction platforms transcends the particular equipment and seems to reside in the recording itself, which may implicate the mic-ing technique, quality, recording chain, etc.
Piano solos seem especially vulnerable to a, for lack of a better description, "gritty" "pumped" sound that is exceedingly annoying. I would guess this may very well be compression induced distortion occurring in the microphone, and is highly audible.
John L.
Hi Andy,
I do not "own" the paper,this site states 0.03% instead of 0.003%. Which number is correct?
Thanks!
I do not "own" the paper,this site states 0.03% instead of 0.003%. Which number is correct?
Thanks!
Re: audibility
Hi John,
The most interesting thing about microphone distortion is that it is quite possible for the distortion at the mic to exceed that of the playback system, given the circumstances you describe above - where the microphone sees very high SPL and the end-listener listens at far lower SPL.
I measured 100dB SPL yesterday on our workshop upright piano - not a very loud piano - measured at 1m. This was an RMS C-weighted measurement, so we can predict at least 15-20dB peaks on top of that.
Given that the average piano CD is recorded with mics often inside the lid of a much louder grand-piano, sometimes with the mics very close to touching the strings, we could easily expect in excess of 130-140dB on the peaks, which means that very few mics in use today would deliver less than 10% distortion (many would deliver closer to 30% distortion at 140dB SPL!).
Orchestral peaks of around 140dB have been measured and a drum kit can do similar damage at close range.
The kind of peak SPLs we see in acoustic music are very far from trivial (and close-mic'ing only makes the problem far far worse).
Andy
PS - one significant caveat might be that the average classical recording has often also been compressed/limited on top of the microphone distortion - so what you hear in the end result is likely to be a combination of issues.
auplater said:
Hi John,
The most interesting thing about microphone distortion is that it is quite possible for the distortion at the mic to exceed that of the playback system, given the circumstances you describe above - where the microphone sees very high SPL and the end-listener listens at far lower SPL.
I measured 100dB SPL yesterday on our workshop upright piano - not a very loud piano - measured at 1m. This was an RMS C-weighted measurement, so we can predict at least 15-20dB peaks on top of that.
Given that the average piano CD is recorded with mics often inside the lid of a much louder grand-piano, sometimes with the mics very close to touching the strings, we could easily expect in excess of 130-140dB on the peaks, which means that very few mics in use today would deliver less than 10% distortion (many would deliver closer to 30% distortion at 140dB SPL!).
Orchestral peaks of around 140dB have been measured and a drum kit can do similar damage at close range.
The kind of peak SPLs we see in acoustic music are very far from trivial (and close-mic'ing only makes the problem far far worse).
Andy
PS - one significant caveat might be that the average classical recording has often also been compressed/limited on top of the microphone distortion - so what you hear in the end result is likely to be a combination of issues.
Hello,
Here is the conclusion of that paper:
These findings can be summarized as follows:
The audible threshold of transient intermodulation distortion was found to be 0,003 % when the RMS distortion is averaged over a period of 250 ms. The corresponding threshold for peak distortion / peak signal ratio is 3 %.
The piano sound was the most sensitive indicator of the distortion.
The most sensitive listener could detect the limitation of the signal first derivative to 4 · 10^3 V/Vs. The most sensitive indicator was pop music. Electrically, completely distortionless reproduction required about 15 · 10^3 V/Vs for the records used.
It does not seen probable that the ear averages distortion stimuli temporally, as assumed earlier. The distortion amplitude itself seems to be relatively unimportant for the sensation of the distortion. The detection of the distortion seems to bear a strong element of linear prediction by the brain; the more unexpected the distortion is and the more harshly it intrudes in the harmonic structure of the actual music, the easier it is to detect.
Best regards from Paris, France.
Jean-Michel Le Cléac'h
Here is the conclusion of that paper:
These findings can be summarized as follows:
The audible threshold of transient intermodulation distortion was found to be 0,003 % when the RMS distortion is averaged over a period of 250 ms. The corresponding threshold for peak distortion / peak signal ratio is 3 %.
The piano sound was the most sensitive indicator of the distortion.
The most sensitive listener could detect the limitation of the signal first derivative to 4 · 10^3 V/Vs. The most sensitive indicator was pop music. Electrically, completely distortionless reproduction required about 15 · 10^3 V/Vs for the records used.
It does not seen probable that the ear averages distortion stimuli temporally, as assumed earlier. The distortion amplitude itself seems to be relatively unimportant for the sensation of the distortion. The detection of the distortion seems to bear a strong element of linear prediction by the brain; the more unexpected the distortion is and the more harshly it intrudes in the harmonic structure of the actual music, the easier it is to detect.
Best regards from Paris, France.
Jean-Michel Le Cléac'h
tiki said:
Re: Re: audibility
Ok I mean no offense, nor do I mean to doubt what you measured, but these numbers are considerably higher than what I have measured myself and seen published by others. Having been involved in setting up and recoding orchestra's in college auditoriums and churches, I too took measurements of the peak spl's during their practice to figure out how far away to place the mic's. During their crescendo's I never measured anything greater than 120db's. I believe that measurement was taken at 1.5 meters, as I ended up going with a 2 meter distance to ensure I didn't overload the mic. In some cases more.
My guess for the disparity is maybe the difference in distance? I mean, I've never tried this, but I would imagine placing a mic at the mouth of one of the brass instruments, or within inches of the Tympani drums might cause such high spl's (they still seem a little high to me), but at the point's where mic's are typically placed, those numbers seem awful high to me.
If those numbers are in fact accurate numbers at 1-2 meters, wouldn't that imply near instant and permanent hearing damage to those performing in the orchestra?
Andy Simpson said:
Ok I mean no offense, nor do I mean to doubt what you measured, but these numbers are considerably higher than what I have measured myself and seen published by others. Having been involved in setting up and recoding orchestra's in college auditoriums and churches, I too took measurements of the peak spl's during their practice to figure out how far away to place the mic's. During their crescendo's I never measured anything greater than 120db's. I believe that measurement was taken at 1.5 meters, as I ended up going with a 2 meter distance to ensure I didn't overload the mic. In some cases more.
My guess for the disparity is maybe the difference in distance? I mean, I've never tried this, but I would imagine placing a mic at the mouth of one of the brass instruments, or within inches of the Tympani drums might cause such high spl's (they still seem a little high to me), but at the point's where mic's are typically placed, those numbers seem awful high to me.
If those numbers are in fact accurate numbers at 1-2 meters, wouldn't that imply near instant and permanent hearing damage to those performing in the orchestra?
general room question with multiple subs
Since my listening room has one large opening to the rest of the house should I be thinking in terms of arranging multiple subs throughout the ground floor or can I just think about the main room?
I use what was designed to be a dining room for my listening room.
Its approx 13 x 14x9 feet high.
The rear wall is open, with a 8x8' opening, so most of that wall is this opening. Beyond this, the space opens into a 7.5' wide 2 story foyer/hallway, and then directly behind this is a mirror image of the room described -so the actual rear wall is about 35' back.
Also there is an open doorway on the left on the listening room.
The rest of the ground floor is almost entirely wide open. The house is "normal sized" I guess, not a huge house by most standards. So I assume I am still in the realm of the small room schroeder frequency, but not sure how much you can open up a room to a larger room before it clouds the way you set this up.
Going with the thinking that at bass frequencies the listener hears the room not the source, what would be the proper way to think of a situation like this?
Also, I currently use a dipole servo controlled sub, and it seems to work pretty well. However, as most things in audio, you never seem to realize deficiency until its in comparison to something better.
My first attempt to add subs:
I tried adding a 2nd, sealed, sub to this but I had a hard time getting them to sound well together. Maybe dipoles and monopoles shouldn't be mixed?
I didn't have the proper tools to measure or do spacial averaging. By ear, I did try several various setting of the phase and gain and locations in the room. Having little more than 3 walls with the mains along one, didn't leave me too many options for placement. I did build a tall stand and tried with it above the centerline, too.
I follow the Linkwitz suggestion to keep the dipole sub angled towards the listener, along the side wall, equidistant to the listener as the mains, which is also near a corner in this case.
Thanks for help.
Tony
Since my listening room has one large opening to the rest of the house should I be thinking in terms of arranging multiple subs throughout the ground floor or can I just think about the main room?
I use what was designed to be a dining room for my listening room.
Its approx 13 x 14x9 feet high.
The rear wall is open, with a 8x8' opening, so most of that wall is this opening. Beyond this, the space opens into a 7.5' wide 2 story foyer/hallway, and then directly behind this is a mirror image of the room described -so the actual rear wall is about 35' back.
Also there is an open doorway on the left on the listening room.
The rest of the ground floor is almost entirely wide open. The house is "normal sized" I guess, not a huge house by most standards. So I assume I am still in the realm of the small room schroeder frequency, but not sure how much you can open up a room to a larger room before it clouds the way you set this up.
Going with the thinking that at bass frequencies the listener hears the room not the source, what would be the proper way to think of a situation like this?
Also, I currently use a dipole servo controlled sub, and it seems to work pretty well. However, as most things in audio, you never seem to realize deficiency until its in comparison to something better.
My first attempt to add subs:
I tried adding a 2nd, sealed, sub to this but I had a hard time getting them to sound well together. Maybe dipoles and monopoles shouldn't be mixed?
I didn't have the proper tools to measure or do spacial averaging. By ear, I did try several various setting of the phase and gain and locations in the room. Having little more than 3 walls with the mains along one, didn't leave me too many options for placement. I did build a tall stand and tried with it above the centerline, too.
I follow the Linkwitz suggestion to keep the dipole sub angled towards the listener, along the side wall, equidistant to the listener as the mains, which is also near a corner in this case.
Thanks for help.
Tony
Tony
Highly couple spaces act like one room, but the subs should be kept closer to the listening space. Your coupled spaces act like one big space, but the energy can flow into one of the other spaces and then back again. This can be a negative effect if the nerby spaces are highly reverberant, but its generally not a problem.
Measurements are almost a must and they are not hard to get. I have even used a Creative Labs mic that was free with a sound card. Works fine at LFs. Software is free and any measurements are going to help.
I've used monopole and dipoles together. Its not an issue, but might be a little harder to set up.
Highly couple spaces act like one room, but the subs should be kept closer to the listening space. Your coupled spaces act like one big space, but the energy can flow into one of the other spaces and then back again. This can be a negative effect if the nerby spaces are highly reverberant, but its generally not a problem.
Measurements are almost a must and they are not hard to get. I have even used a Creative Labs mic that was free with a sound card. Works fine at LFs. Software is free and any measurements are going to help.
I've used monopole and dipoles together. Its not an issue, but might be a little harder to set up.
My room also is built with a somewhat open architecture, and I've been using a variation of Dr. Geddes approach, and have achieved smooth bass response.
On a related note, would small partitions filled with fiberglass insulation act as a sort of bass trap? I mean, I have a sort of bar area separating my kitchen from my living room. For who knows what reason, the wall is drywalled on both sides, but their appears to be fiberglass insulation in their (I can see it poking out where the pipes come in). I also have a closet which is basically an enclosed area that does not go all the way to the ceiling, it's drywalled all around, as well as inside, and I've wondered if this cavity may also act as a low frequency absorber, as it seems pretty similar to the execution of many bass absorbers.
On a related note, would small partitions filled with fiberglass insulation act as a sort of bass trap? I mean, I have a sort of bar area separating my kitchen from my living room. For who knows what reason, the wall is drywalled on both sides, but their appears to be fiberglass insulation in their (I can see it poking out where the pipes come in). I also have a closet which is basically an enclosed area that does not go all the way to the ceiling, it's drywalled all around, as well as inside, and I've wondered if this cavity may also act as a low frequency absorber, as it seems pretty similar to the execution of many bass absorbers.
I once cut a partition down to its studs, filled it with fiberglass and covered it over with a felted type of material. It was a very effective absorber and looked just like a wall with wallpaper on it. Only problem was that it was not rigid and if bumped or leaned on it would show the impression. This may be curable.
Jmmlc said:
Interesting that the piano was found 'most sensitive'.
If we knew what microphones had been used to record the test stimuli and in what manner, we might be able to predict the significance of the microphone non-linearities and possibly pose the question of significance to the test.
For example, the piano may have fared best because it is a lower SPL source than a symphonic orchestra, which would mean that if the microphones & distance were constant, the distortion at the microphone would be far lower.
Whether we can say that this would bias the test I'm not sure.
Given the date that the paper was written, we can reasonably predict that if the recordings were 'standard studio recordings', the level of distortion in the microphones available for the symphonic orchestral peaks could have been anything from ~1% to >10% (if not close-mic'd) and an order of magnitude (or more) worse if close-mic'd.
Further, we must also consider the playback SPL of the test and the distortion of the output transducer system.
Taking a potentially grossly distorted recording as a starting point, I wonder how we might expect relatively 'similar/small amounts' of distortion to be audible?
Certainly, if we were testing noise-floor, the noise-floor of the source recordings would be totally critical to the measured threshold of audibility, if the levels were comparable (as in this case).
Andy
PS - if the moderator wishes to separate the discussion from the thread I have no objection.
Re: Re: Re: audibility
To clarify - do I assume you measured with a calibrated microphone or with an 'SPL meter'?
Very few meters actually provide real peak metering and while they will show 'the highest RMS level' (often called 'peak'), this is not really the actual peak level but the loudest average.
This type of meter is the standard way of measuring the loudness of an event but we have to look well above this measurement for the real peaks.
Assuming that your measurements were made with a calibrated mic;-
There are a few factors here that might not at first glance seem important.
Firstly, it is not necessarily crescendos but PEAKs that we are interested in.
Secondly, given that we are talking about peaks, the group itself is absolutely critical to this.
I don't suggest that the 120dB orchestra in question was bad but at the college level two things must be taken into account.
Firstly, the college orchestra does not have the technique of the pro orchestra, so we can expect their individual output levels to be lower (they have less experience in 'projecting' their sound - or simply are playing less efficiently).
Secondly, the college orchestra does not have the experience/timing of the pro orchestra, which means that where peaks occur individually, they are very much less likely to occur close-enough in time to achieve greater peaks than are possible alone.
In other words, while the RMS loudness of the college orchestra might seem similar to that of a good pro orchestra, the peaks of the pro orchestra I would expect to be significantly higher.
Yes indeed. Many a conductor has prematurely lost his hearing.
This is perhaps not surprising though, given that he is the focal point in time AND space of the entire orchestra, so when those peaks happen, he makes sure that they should be aligned in time most perfectly at his location!
In any case, I will try to recall the reference from which I took the quoted numbers (similar numbers can be seen on the 'hear.net' site, if I recall).
Andy
pjpoes said:
To clarify - do I assume you measured with a calibrated microphone or with an 'SPL meter'?
Very few meters actually provide real peak metering and while they will show 'the highest RMS level' (often called 'peak'), this is not really the actual peak level but the loudest average.
This type of meter is the standard way of measuring the loudness of an event but we have to look well above this measurement for the real peaks.
Assuming that your measurements were made with a calibrated mic;-
There are a few factors here that might not at first glance seem important.
Firstly, it is not necessarily crescendos but PEAKs that we are interested in.
Secondly, given that we are talking about peaks, the group itself is absolutely critical to this.
I don't suggest that the 120dB orchestra in question was bad but at the college level two things must be taken into account.
Firstly, the college orchestra does not have the technique of the pro orchestra, so we can expect their individual output levels to be lower (they have less experience in 'projecting' their sound - or simply are playing less efficiently).
Secondly, the college orchestra does not have the experience/timing of the pro orchestra, which means that where peaks occur individually, they are very much less likely to occur close-enough in time to achieve greater peaks than are possible alone.
In other words, while the RMS loudness of the college orchestra might seem similar to that of a good pro orchestra, the peaks of the pro orchestra I would expect to be significantly higher.
Yes indeed. Many a conductor has prematurely lost his hearing.
This is perhaps not surprising though, given that he is the focal point in time AND space of the entire orchestra, so when those peaks happen, he makes sure that they should be aligned in time most perfectly at his location!
In any case, I will try to recall the reference from which I took the quoted numbers (similar numbers can be seen on the 'hear.net' site, if I recall).
Andy