What is this "Noise" so many Audiophiles are talking about? Anyone really Understand?
- Thread starter caesar
- Start date
Tao,
Thanks. You are definitely correct. I think a lot about making my system better. Yet I don't see myself as an audiophile because I don't really buy into a lot of the orthodoxies of the audiophile hobby. I kind of see myself more of a guy who enjoys a luxury experience of listening to my system 10-12 hours a week. Because I have so many things going on in life, and this is just one hobby of many, I step away from it and bring an outsider's perspective. I also have deep background in applied economics and behavioral sciences, so I see the hobby, including the marketing and persuasion, through that lens.
A more accurate description of my position to the audiophile world is Georg Simmel's construct of the Stranger - as I straddle the outskirts of the audiophile social network. Some of the reasons for not getting "more inside" include:
- I see the audiophile culture as kind of bankrupt. It is a subjective hobby, yet experiences aren't properly shared
- The elites and reviewers are mis-incentivized and not properly educated on subjectivism, so they spread a lot of false information, making people waste time and money auditioning gear that doesn't fit their taste
- Many of the elites act is if they have superior, mystical senses because they imagine certain things inside their minds better than someone else, as evidenced by the recent physical confrontation between "Great" Peter Breuninger and "Sterile" Jon Valin about the "best turntable"
- many audiophiles play the "Plato's cave game" and compare their system to actual live performances, which, being a realist and an Aristotelian, I find silly
- many audiophiles love to argue that their taste is better than someone else's, hence all the tubes vs. ss, digital vs. vinyl debates, etc., when all types of technologies put people in a state of musical flow and bliss
- Most of the systems I have heard suck. The better ones may do some things well, such as the see-through transparency of a Martin Logan, resolution of a Magoco, dynamics of a Wilson or a Vivid, expression of the midrange by a horn, tonality of great tube gear, etc., but I have experienced very few systems that are truly magical
... I can go on and on...
The audio "noise" I have heard is almost always when the system is played at louder levels and things descend into a hash of sound..at lower levels , the audible noise is almost always because of extraneous noise in the listening area
Empirical Audio
Industry Expert
Two types of noise in audio systems:
1) background noise that is always present, hiss, hum etc..
2) noise that superimposes or "rides" on the music waveform - it's not audible until you play a track
#2 is harder to identify and diagnose, but it's very real.
Steve N.
Empirical Audio
1) background noise that is always present, hiss, hum etc..
2) noise that superimposes or "rides" on the music waveform - it's not audible until you play a track
#2 is harder to identify and diagnose, but it's very real.
Steve N.
Empirical Audio
I believe this is what most audiophiles are referring to when they talk about "noise" (electrical). If digital transmission is a modulation of analog such as voltage - There is no such thing as Digital https://www.audiostream.com/content/draft?page=1,
- many argue that electrical noise can either hitch along for the ride affecting downstream reception and/or be incorporated into and alter the audio signal.
David
Well said.
#2 is a big one. You'll notice it the most when it's gone after system improvements.
#1 can be distracting at times, but it is less detrimental than #2 since it doesn't really modulate the music, is unrelated to it.
Wow, where does one start replying to this?!
I don't think the allegory of Plato's cave is fundamentally different to Aristotlean realism only in its explanation of how perceptions might be illusory. Plato describing them as shadows on the cave wall (and better explained by concepts of the absolute mind, spirit or soul) and realists believing object reality exists independent of the human mind (and better explained by scientific inquiry).
The Aristotelian view and scientific method is still based on observations. In music perception those observations rely on the subjective senses. Of course mathematical observations and measurements of data are more "objective". The question always is are we measuring what we think we are. As Einstein used to say (quoting from William Cameron apparently) “Not everything that can be counted counts, and not everything that counts can be counted.” The audio signal and sound wave and neural signal remain surrogates, indirect markers, of what is the end game - music perception in the brain. Until that point it simply isnt music. And it is more than just a passive registration of a signal on the cortex.
Then there is view that reality is malleable, changing with fresh evidence of objects and things. How we know reality is from scientific inquiry when that inquiry all goes well. The pursuit of truth through inquiry and collection of evidence is therefore fallible.
I dont agree that audio reviewers "spread a lot of false information, making people waste time and money auditioning gear that doesn't fit their taste". They are reporting their subjective experiences. Whether it aligns with yours or whether it suits your taste is up to you to decide when you get to hear the same gear. Reviews are useful IMO in sorting what to audition once you get a sense of which reviewer more closely matches your experience. Of course any reviewers' account can be biased by his/her tastes or other factors and it is also true tastes can be acquired.
Cheers
David
Is there any evidence that the noise can "ride" on the music waveform? That implies that the noise will vary with the signal.
What seems more likely to me is that the music "rides" on the noise and the noise has a stochastic influence on the music signals.
I'm not sure the distinction between the two matter as you can see noise as it's own waveform or summed with the music waveform. Both are correct afaik. But the noise isn't audible and so it doesn't technically change what you'd actually hear... it's not audible. But it does, it has some negative side effects unrelated to how the two waveforms sum.
The noise's effect on the signal is an interesting topic. I don't think it's random, it's effect is going to depend on changes in voltage and current with respect to time, maybe other things. We've all heard the effect of reducing noise in a system, it's audibly definable and with experience easy to identify by ear. So I'd think noise's effect on the music signal is going to be a very complex phenomenon.
Yes, Norman Crowhurst and others have commented on this topic and that it is a consquence of using negative feedback where you end up with a myriad of distortion components that in fact creates a signal modulated "noise" floor. The problem with this is that, being correlated with the music signal, it will mask low level signals whereas true noise will not (to a limit of course) to the same degree because it is uncorrelated to the signal. Your brain treats correlated and uncorrelated signals very differently.
There are in fact many types of "noise" that are either not directly related to the signal or are correlated with the signal.
There is thermal noise, which is essentially white noise and cumulatively can be heard as the "hiss" one hears from many electronics. This is usually simple noise and not generally correlated with the signal in any way.
There is hum, which can be mechanical, electrical or both. This is seen as harmonics from the AC line frequency (60, 120, 180, 240 etc. in the US and 50, 100, 150, 200 etc. in Europe). This can be either simple noise and uncorrelated with the signal OR it can be correlated with the signal in some circuit designs. This can be clearly seen in an FFT scan where the harmonic distortion components are surrounded by a lot of "fuzz", which is modulation of the harmonic distortion components with the frequencies generated from the line voltage. So, from noise we get a new kind of distortion on the signal. Here is a clear indication of where power cleanliness is potentially very important to the sound we hear.
Look at this link to see what I mean by modulation of the harmonic distortion with power supply harmonics.
https://www.stereophile.com/content...n-mono-monoblock-power-amplifier-measurements
What you can see is that there is a regular series of line harmonics in the noise floor (Figure 3)
There is clear evidence of modulation in the 50hz FFT (Figure 8)
There is clear evidence of modulation in the 19+20Khz IMD FFT (Figures 9 and 10) below 5Khz...it is more evident at 1 watt where the high order harmonics have dropped to the noise floor. There is also a lot of fuzz around the 19 and 20 Khz peaks as well as the sidebands at 100 watts.
https://www.stereophile.com/content/audio-alchemy-dpa-1m-monoblock-power-amplifier-measurements
This one also shows intermodulation with the power supply harmonics.
There are even worse examples but I thought an example from a couple very modern designs would highlight that this is far from a solved problem in most amps.
Then there is RFI and EMI, these can also work there way into the signal chain and although I have not seen measurements showing their impact (like the line noise example) on audio, I have seen the effects first hand working with Mass Spectrometers and laser systems. Our laser system was a pulsed Nd:YAG system that had some ultra high speed switches. These switches and the high voltages controlled by them generated huge bursts of RFI that we were picking up on the coaxial (shielded of course) signal cables. Clearly the shielding they had was not adequate, which is why we were wrapping critical areas with aluminum foil and then running grounding wires from that foil. It didn't eliminate the problem but reduced it. So, I know these parasitic RFI and EMI signals can pollute your noise floor but even worse is when these things get modulated with the signal (like the hum example) and create more "noise", which is in fact a new kind of distortion.
There is also back EMF "noise", which is really a form of distortion but will only afflict amplifiers with substantial amounts of negative feedback. This was first pointed out by Matti Otala in a couple of papers in the 1980s. This is another form of signal correlated noise that then gets intermodulated with the music signal thus further reducing fidelity.
Finally there is negative feedback itself. Negative feedback reduces the levels of lower order harmonics and results in the generation of high order harmonics. This has been demonstrated by Baxandall, Crowhurst, Pass and others. The argument is that THD is reduced but as Crowhurst pointed out you get an endless array of high order harmonics, multiples of multiples of multiples. This creates a complex "noise" floor that is signal correlated and modulated. This means it is moving up and down. It is fundamentally different than the thermal noise or unmodulated hum noise, which are not correlated with the signal and you can actually hear correlated content BELOW a true random noise floor...what you cannot do is hear those very soft signals below a non-random noise floor because the correlation matters to our ear/brain pattern recognition.
IMO, what people talk about when there is no music playing and what they mean by a lowered noise floor when the music is playing are two quite different things. The noise with no music playing is the intrinsic noise in the system from thermal and hum maybe some RFI/EMI effects as well. However, when the music plays, you have potential intermodulation with all the noises listed above except perhaps the white thermal noise (maybe there as well). How well a circuit design rejects interferrence from power supply noise, RFI/EMI, back EMF will have a big impact on how low it's "noise" floor is and esepcially if there is a lot of negative feedback involved, which will make many of these intermodulation effects more rather than less likely.
I think when people hear a big benefit from a power cable that is good at rejecting RFI/EMI, it indicates that their electronics were being negatively impacted and a percpetion of a "blacker background" is one of the most commone benefits. I also think that this is one of the reasons good SET amps, if not pushed too hard, often do a better job of making a 3d image/soundstage because they are not being hampered by the negative feedback and back EMF issues (although they most certainly can have issues with the power supply intermodulation and RFI/EMI contamination).
There is thermal noise, which is essentially white noise and cumulatively can be heard as the "hiss" one hears from many electronics. This is usually simple noise and not generally correlated with the signal in any way.
There is hum, which can be mechanical, electrical or both. This is seen as harmonics from the AC line frequency (60, 120, 180, 240 etc. in the US and 50, 100, 150, 200 etc. in Europe). This can be either simple noise and uncorrelated with the signal OR it can be correlated with the signal in some circuit designs. This can be clearly seen in an FFT scan where the harmonic distortion components are surrounded by a lot of "fuzz", which is modulation of the harmonic distortion components with the frequencies generated from the line voltage. So, from noise we get a new kind of distortion on the signal. Here is a clear indication of where power cleanliness is potentially very important to the sound we hear.
Look at this link to see what I mean by modulation of the harmonic distortion with power supply harmonics.
https://www.stereophile.com/content...n-mono-monoblock-power-amplifier-measurements
What you can see is that there is a regular series of line harmonics in the noise floor (Figure 3)
There is clear evidence of modulation in the 50hz FFT (Figure 8)
There is clear evidence of modulation in the 19+20Khz IMD FFT (Figures 9 and 10) below 5Khz...it is more evident at 1 watt where the high order harmonics have dropped to the noise floor. There is also a lot of fuzz around the 19 and 20 Khz peaks as well as the sidebands at 100 watts.
https://www.stereophile.com/content/audio-alchemy-dpa-1m-monoblock-power-amplifier-measurements
This one also shows intermodulation with the power supply harmonics.
There are even worse examples but I thought an example from a couple very modern designs would highlight that this is far from a solved problem in most amps.
Then there is RFI and EMI, these can also work there way into the signal chain and although I have not seen measurements showing their impact (like the line noise example) on audio, I have seen the effects first hand working with Mass Spectrometers and laser systems. Our laser system was a pulsed Nd:YAG system that had some ultra high speed switches. These switches and the high voltages controlled by them generated huge bursts of RFI that we were picking up on the coaxial (shielded of course) signal cables. Clearly the shielding they had was not adequate, which is why we were wrapping critical areas with aluminum foil and then running grounding wires from that foil. It didn't eliminate the problem but reduced it. So, I know these parasitic RFI and EMI signals can pollute your noise floor but even worse is when these things get modulated with the signal (like the hum example) and create more "noise", which is in fact a new kind of distortion.
There is also back EMF "noise", which is really a form of distortion but will only afflict amplifiers with substantial amounts of negative feedback. This was first pointed out by Matti Otala in a couple of papers in the 1980s. This is another form of signal correlated noise that then gets intermodulated with the music signal thus further reducing fidelity.
Finally there is negative feedback itself. Negative feedback reduces the levels of lower order harmonics and results in the generation of high order harmonics. This has been demonstrated by Baxandall, Crowhurst, Pass and others. The argument is that THD is reduced but as Crowhurst pointed out you get an endless array of high order harmonics, multiples of multiples of multiples. This creates a complex "noise" floor that is signal correlated and modulated. This means it is moving up and down. It is fundamentally different than the thermal noise or unmodulated hum noise, which are not correlated with the signal and you can actually hear correlated content BELOW a true random noise floor...what you cannot do is hear those very soft signals below a non-random noise floor because the correlation matters to our ear/brain pattern recognition.
IMO, what people talk about when there is no music playing and what they mean by a lowered noise floor when the music is playing are two quite different things. The noise with no music playing is the intrinsic noise in the system from thermal and hum maybe some RFI/EMI effects as well. However, when the music plays, you have potential intermodulation with all the noises listed above except perhaps the white thermal noise (maybe there as well). How well a circuit design rejects interferrence from power supply noise, RFI/EMI, back EMF will have a big impact on how low it's "noise" floor is and esepcially if there is a lot of negative feedback involved, which will make many of these intermodulation effects more rather than less likely.
I think when people hear a big benefit from a power cable that is good at rejecting RFI/EMI, it indicates that their electronics were being negatively impacted and a percpetion of a "blacker background" is one of the most commone benefits. I also think that this is one of the reasons good SET amps, if not pushed too hard, often do a better job of making a 3d image/soundstage because they are not being hampered by the negative feedback and back EMF issues (although they most certainly can have issues with the power supply intermodulation and RFI/EMI contamination).
There is interaction between the two, even if the noise content is random. The interaction can result in a stochastic resonance in which sub-noise threshold audio can be made audible and supra-threshold signals can be amplified if they fall into the noise frequency bandwidth.
Empirical Audio
Industry Expert
Waveforms passing through resistors cause thermal noise. Is this distortion or noise??
Most of the time, distortion is related to the waveform interacting with the electronics.
Noise on the other hand can be related to the waveform interacting with electronics and wiring, as well as uncorrelated noise, which is present even with no waveform active.
Steve N.
Empirical Audio
Generally you're right. But the frequency of the noise is a dependent variable. Depending on whether the noise comes is the other variable.
High frequency RF is so problematic that to counter the effects at the point of say a transistor, actually presents some challenges. In that situation it's very easy & typical for the two to become one in the same, even though the noise could be coming from the signal source. Something to point out is that devices that have a bandwidth that say stops at 45mhz, don't stop anything higher than that from going through them or interacting. The point is that after 45mhz they can't accurately control the higher frequencies, and parasitics can take over. Parasitics like how much capacitance they have, lending aid to RF jumping around, etc.
Now I claimed "problems" but the reality is countless manufacturers and countless audiophiles outright enjoy the benefit of added noise, in the right frequency spectrum. Generally speaking noise that's low in the RF or in the audible range is bad for clarity in the music. But RF at some point actually brings clarity in the sense that it drives details forward. It's artificial, and not my preference, but some people seriously enjoy hearing what what was dominated by the fundamentals of the music prior.
You can force feedback into many more high order harmonics, but you don't really see amp designers doing that much these days. Generally speaking they are pretty good at it and the harmonics peter-out after the 3rd, in a hurry. It's important to note that you're not increasing the distortion level, you're just rearranging it. But the higher the frequency the less it was reduced, so it can look as though you increased some distortion, even though the reality is it just didn't go down as much.
My system is extremely quiet. I don't hear anything when I stand next to the speakers with no music playing.
BY FAR, the most noise (albeit muffled) comes from outdoors and other people breathing in the house.
BY FAR, the most noise (albeit muffled) comes from outdoors and other people breathing in the house.
Great post up above from Brad, the breakdown of different electrical noises was interesting and comments in general on feedback as a noise as well.The amounts of feedback and whether it is global is always something I look for in amps and definitely a less is more approach often seems to be a good one.
If you live in a constantly noisy environment and then have gear that is generally obviously noisy with hum or fan noises, or if you have horns at 100+ db sensitivity I suppose at some level your perception seems to give in and tend to filter and cancel out some of the components of this noise floor and it becomes a bit like very mild tinnitus where you try and manage and listen through by focus. What this requires of our brains and if this comes at a cost or teaches us how to focus better I have no idea really but generally desensitising is helpful to some degree if there was a hypersensitivity to these things in the first place but certainly too much desensitising comes at some cost.
The differences between noise and distortion can maybe also come down to more about definitions and context. In the end noises and distortions are only an issue if you don’t like them. If you view these things holistically both mask the true signal which for me is the performance rather than the recording. There are clearly going to be distortions all along the chain from recording through to replay and finally in the perception. Noises can occur anywhere along the chain and also be captured in the recording.
For me noise and distortion both mask signal. There are more audible levels of each and also levels where they are barely audible but still have significant bearing on how we perceive the true signal... the performance and the music.
If you live in a constantly noisy environment and then have gear that is generally obviously noisy with hum or fan noises, or if you have horns at 100+ db sensitivity I suppose at some level your perception seems to give in and tend to filter and cancel out some of the components of this noise floor and it becomes a bit like very mild tinnitus where you try and manage and listen through by focus. What this requires of our brains and if this comes at a cost or teaches us how to focus better I have no idea really but generally desensitising is helpful to some degree if there was a hypersensitivity to these things in the first place but certainly too much desensitising comes at some cost.
The differences between noise and distortion can maybe also come down to more about definitions and context. In the end noises and distortions are only an issue if you don’t like them. If you view these things holistically both mask the true signal which for me is the performance rather than the recording. There are clearly going to be distortions all along the chain from recording through to replay and finally in the perception. Noises can occur anywhere along the chain and also be captured in the recording.
For me noise and distortion both mask signal. There are more audible levels of each and also levels where they are barely audible but still have significant bearing on how we perceive the true signal... the performance and the music.
Great post from Morricab et al.
If removing/reducing noise is the goal, given the different types, it would suggest a multi-pronged approach.
The big issue (??) seems to be stripping out the signal (or at least current) along with the noise - doing more harm than good. Gryphon repeatedly suggest not to use power conditioners and the like.
Solar power is becoming more ubiquitous these days which raises other questions. Are the PV panels polluting the grid with harmonics? If you have your own PV system, does an inverter add more noise than normally expected via the grid?
David
If removing/reducing noise is the goal, given the different types, it would suggest a multi-pronged approach.
The big issue (??) seems to be stripping out the signal (or at least current) along with the noise - doing more harm than good. Gryphon repeatedly suggest not to use power conditioners and the like.
Solar power is becoming more ubiquitous these days which raises other questions. Are the PV panels polluting the grid with harmonics? If you have your own PV system, does an inverter add more noise than normally expected via the grid?
David
Great post Morricab (Brad?) - a great outline of the various noise forms & their results.
Totally agree that noise which modulates with signal is probably the most insidious of the noise categories & probably the last one to be identified.
Also totally agree that modulated noise is perceived far differently to static noise. This difference, I believe, is down to one of the quirks of auditory perceptions which has a difficult name "co-modulation masking release" (CMR). It describes a phenomena that we can hear a signal which is below the noise floor when there are signals at other frequencies which are modulating in unison with the buried signal. This is a common feature of natural sounds & it appears that our auditory processing has evolved to take advantage of this fact & to be able to hear sounds which would normally be inaudible as they are buried beneath the noise floor. The actual mechanism underlying this ability of CMR to hear beneath the noise floor is still not resolved even though the effect has been know for a very long time.
One of my premise is that noise floor modulation which is generated in the electronics & is signal related, creates a non-natural perception of the sound in some way. I suspect that the correlated noise created in the electronics is not of the same form as found in natural soundscapes but our auditory processing still latches onto it (as it is the way auditory processing works) & as a result, is confused by it?
Totally agree that noise which modulates with signal is probably the most insidious of the noise categories & probably the last one to be identified.
Also totally agree that modulated noise is perceived far differently to static noise. This difference, I believe, is down to one of the quirks of auditory perceptions which has a difficult name "co-modulation masking release" (CMR). It describes a phenomena that we can hear a signal which is below the noise floor when there are signals at other frequencies which are modulating in unison with the buried signal. This is a common feature of natural sounds & it appears that our auditory processing has evolved to take advantage of this fact & to be able to hear sounds which would normally be inaudible as they are buried beneath the noise floor. The actual mechanism underlying this ability of CMR to hear beneath the noise floor is still not resolved even though the effect has been know for a very long time.
One of my premise is that noise floor modulation which is generated in the electronics & is signal related, creates a non-natural perception of the sound in some way. I suspect that the correlated noise created in the electronics is not of the same form as found in natural soundscapes but our auditory processing still latches onto it (as it is the way auditory processing works) & as a result, is confused by it?
Last edited:
Absolutely. Think microphonics and vibration issues with a piece of gear with a tube(s) Kal. The unwanted artifact will most definitely "ride" the signal.
Tom
Actually looking at data published by Stereophile and Soundstage does not bear out what you are saying. Many modern amps today have significant high order distortion components and not only at higher frequencies like 1Khz or the 19+20Khz IMD test but also at 50Hz. It is the nature of feedback that this is so and has nothing to do with the deisgner's wish. It is clear that the total THD is reduced but the order is increased and based on the work starting back in the 1940s and 50s it is also clear that higher order harmonics are far more detrimental to the sound than 2nd order. Also, as Jena Hiraga and Cheever have pointed out, the pattern is also important and that it be monotonic, ie. exponentially decreasing with increasing harmonic order. The rearrangement, as you put it, IS the problem. The myriad of components making the new "noise" floor is also a big problem with low level resolution and spatial representation.
What it does with IMD is also interesting and distressing.
Similar threads
