Can You Trust Your Ears? By Tom Nousaine

[fas42]

The 'quality' in sound reproduction is quite easy to assess, subjectively - if one wishes the sound system to be an effects unit, or "put on a performance" which is particularly spectacular on certain recordings that's a different matter; and then preference will most assuredly play a part. IME competent playback will often sound disappointing to those who want a show - because they're used to exaggerated sound characteristics, and this will be missing on a capable system.

Natural sounds can be intense, and one can be comfortable in the midst of a veritable cacophany of aural input of such - being able to create the same subjective experience from playback is the mark of quality; not the precisely controlled, "precision" rendition of highly specific recordings.

 

[Fitzcaraldo215]

It seems to me that human hearing is quite a lot like vision: it is able to focus on the source of the sound and ignore the extraneous stuff. A human can tell pretty well if a TV picture is neutral, and certainly the way my hearing seems to work, can tell a neutral-ish, natural-ish source of sound regardless of the surroundings - even if they have to move their head a little to confirm it.

The analogy with image-based measurements is that if I stick a camera without a lens at the human viewer's position in a room furnished in arbitrary colours, it isn't possible to tell from measurements whether a distant picture is neutral; this, I would say, is the level at which typical acoustic measurements are made with a mic. But if we fit a lens and interpret the image correctly, the source can be distinguished from everything else - which is what I would guess our hearing does to some extent.

Quoting Floyd Toole (but not accusing him of agreeing with me about the how and the why):


If there is anything to this, then there may be some exaggeration in the notion that "all sound is coloured". Just because a 'dumb' measurement with a microphone shows a certain colouration, doesn't mean that human hearing is going to perceive it that way.

I think you are using Toole's views in a narrower, out of context way than he would agree to. Yes, we definitely adapt somewhat to the sound we have, but we might prefer a different sound, which is based on a better sonic model or standards, IF we only had a chance to hear that in comparison. If we have not heard that comparison, then ignorance is bliss!

 

[j_j]

Actually, all stereo systems put up an illusion. You get to choose your illusion. Direct/indirect ratio varies by listener. Frequency balance between direct and playback room reverb varies by listener. Diffusion of playback room, ditto. Liveness or deadness of playback room, ditto.

Once you're reduced to stereo, it's preference all the way.

 

[Cosmik]

I am a bit skeptical of Sean Olive's conclusions on that subject. What does flat frequency response in a sound system give you with a recording made with microphones that do not have flat frequency responses? When one looks at the vast body of commercial recordings over the past hundred years or so how many were recorded only with mics that had a flat frequency response for 20-20kHz? I'm going to go with pretty close to zero out of god knows how many recordings we are talking about. So what do you actually get? You get an accurate representation of colored microphones and subjectiove choices by recording engineers about microphone placement and subjective choices about final mixes and any post processing that may have been done.

I think we are pretty good at assessing the colour balance of computer monitors (that generate their own illumination) and also printed photographs. We can even tell when a colour-biased image is being displayed on a neutral TV and vice versa as long as it is changing dynamically, because of cues and clues that arise in the content.

We can do it by a comparison against familiar objects and illumination, and I think we also have a 'pseudo-absolute' sense of colour that adapts only slowly - people in the UK will have experienced the 'red sun' recently, and even after half an hour or so outside, it still looked very weird; I have no doubt that after a longer period it would have begun to look normal, though.

It feels to me the same with sound; that with continuous exposure to a certain type of coloured sound it would eventually begin to sound normal, but that as long as it is not continuous, and we are frequently exposed to natural sound not just coloured audio, our 'sound colour detector' is continuously being reset to neutral. As with TV screens, we can distinguish between coloured recordings on a neutral system and a continuously coloured system. Tie that to Toole's "hearing through the room" to the source and I think that a 'neutral' system is pretty obvious to humans.

 

[Analog Scott]

I think we are pretty good at assessing the colour balance of computer monitors (that generate their own illumination) and also printed photographs. We can even tell when a colour-biased image is being displayed on a neutral TV and vice versa as long as it is changing dynamically, because of cues and clues that arise in the content.

We can do it by a comparison against familiar objects and illumination, and I think we also have a 'pseudo-absolute' sense of colour that adapts only slowly - people in the UK will have experienced the 'red sun' recently, and even after half an hour or so outside, it still looked very weird; I have no doubt that after a longer period it would have begun to look normal, though.

It feels to me the same with sound; that with continuous exposure to a certain type of coloured sound it would eventually begin to sound normal, but that as long as it is not continuous, and we are frequently exposed to natural sound not just coloured audio, our 'sound colour detector' is continuously being reset to neutral. As with TV screens, we can distinguish between coloured recordings on a neutral system and a continuously coloured system. Tie that to Toole's "hearing through the room" to the source and I think that a 'neutral' system is pretty obvious to humans.

Color is a very tricky thing. Every monitor has it's limitations as does every digital file. Getting color profiles in line can be a nightmare between different devices. Getting what you see on a monitor to match a print is actually pretty much impossible. The two media are too different in nature for that to happen. A lot of careful tweeking is often needed to even get it close. Also we have the issue of white balance. Our eyes are constantly adjusting our internal white balance. We don't see color the same way moment to moment the way film stock or digital sensors do. If you are comparing a printed image to one on a computer monitor the actual color balance of the light in the room will have a huge effect on the print. If you capture an image of something out doors using a true white balance the image will look very blue on your monitor or print. Liekwise it will look very orange if taken indoors under tungston balanced light. Film stocks are rated for daylight or tungston. Digital cameras have white balance adjustment controls. but what one is actually doing with these things is literally coloring the image to adjust it to our ever adjusting internal white balance in our brains. True accurate color in print, on a computer display or film projection rarely looks right to us. You don't find true balanced white light in nature and rarely in artificial light. Although now with color temperature adjustable LED lights. You actually can get true white light.

With sound we don't have the same thing as we do with vision in that our ears are fixed. We don't have the equivalent of a constantly adjusting apature and white balance in our ears. all we have is the ability to move our heads to face the sound source so as to hear it more clearly.

I'm not sure I think one can break sound down into catagories of "natural" and "colored." Particularly in audio. It's all colored in audio. There is no such thing as an uncolored microphone. And even if there were it's still only capturing sound at a subjectively chosen position with all the limitations of it's directionality. And even in real life sound varies tremendously. I'd be curious to read Toole's "hearing through the room." But hearing to the source? The source being what? A recording that used colored microphones that were placed somewhere in the room based on the subjective judgement of some recording engineer? How does one get "uncolored" sound in audio? And what exactly is "uncolored" or "natural" sound in real life? There s no standardization for sound in real life. Every musical instrument, every musician, every concert hall all have unique sound qualities. Two different people can go to the same exact live concert and hear two completely different shows in terms of sound quality just based on where their seats are in the concert hall. There simply are no fixed standards for the sound quality of music.

 

[Analog Scott]

Actually, all stereo systems put up an illusion. You get to choose your illusion. Direct/indirect ratio varies by listener. Frequency balance between direct and playback room reverb varies by listener. Diffusion of playback room, ditto. Liveness or deadness of playback room, ditto.

Once you're reduced to stereo, it's preference all the way.

I would add 10 likes to this post if I could.

Which only reminds just how lame internet forums have made us. I got 15 likes today on facebook. Only 15! My life sucks. (please like this post)

 

[Wombat]

Current technology can measure signal fidelity through the reproducing system from recorded input up to the ears with a high level of precision. Comparison between systems electrical/mechanical/acoustic performance can be compared. Research has been conducted on the rest of the process, with various aims in mind.
Even with trained listeners there will be individual responses in each of the processes. The diagram shows non-audio and psychological inputs and feedback loops. The affect of these can be significant and unnoticed by an active listener. Mood, bias/expectation, emotional response, environment, preference, involvement of others, and more, can be added to the 'other high-level processes' stage.

Thus, I hear what I perceive but not necessarily what someone else perceives.


For those who aren't familiar with the McGurk effect:

 

[Analog Scott]

"Current technology can measure signal fidelity through the reproducing system from recorded input up to the ears with a high level of precision. Comparison between systems electrical/mechanical/acoustic performance can be compared Research has been conducted on the rest of the process, with various aims in mind."




I think you took the chain too far. I think the correct statement would be "Current technology can measure signal fidelity through the reproducing system from recorded input up to the speaker terminals with a high level of precision." Once the speakers and room are in play technically measured "fidelity" of the "recorded input" becomes a mess. You can't really measure the complex 3 dimensional acoustic waveform that you get in the playback room against the electrical signals at the speaker terminals. I'm not saying you can't measure speakers or assess certain aspects of speaker performance in terms of accuracy. But there are aspects of speaker output and room interaction that simply can not be measured in terms of accuracy to the source signal. Audio signals have no such things as radiation patterns. There is no such thing as a speaker radiation pattern that is more or less accurate in comparison to the source signal.

 

[RayDunzl]

I think we are pretty good at assessing the colour balance of computer monitors

I may not be.

I use f.Lux which adjusts the color balance between day and night settings.

Day looks good during daytime and very blue if selected at night.

Night looks good at nighttime and very red if selected during the day.

 

[Analog Scott]

I may not be.

I use f.Lux which adjusts the color balance between day and night settings.

Day looks good during daytime and very blue if selected at night.

Night looks good at nighttime and very red if selected during the day.

And so you fall into the catagory of normal....when it comes to color perception.

 

[fas42]

I think we are pretty good at assessing the colour balance of computer monitors (that generate their own illumination) and also printed photographs. We can even tell when a colour-biased image is being displayed on a neutral TV and vice versa as long as it is changing dynamically, because of cues and clues that arise in the content.

We can do it by a comparison against familiar objects and illumination, and I think we also have a 'pseudo-absolute' sense of colour that adapts only slowly - people in the UK will have experienced the 'red sun' recently, and even after half an hour or so outside, it still looked very weird; I have no doubt that after a longer period it would have begun to look normal, though.

The rendition of faces, and the outside environment are the things to go by - when the clip is a news report, or recording of an event. No fiddling is done, normally - and a match with the trees, shrubs, sky outside will be spot on - if calibration of one's set has been done.

This means that morning shows have pinkified faces, lots of orange faces in station promos, quite washed out colour in dramas intending to evoke 'old times'; recent Brit series have pumped up the colour, for some reason - on balance, the colours "feel" like they're correct for the situation, and the known "style" of the producers of the shows.

 

[Analog Scott]

The rendition of faces, and the outside environment are the things to go by - when the clip is a news report, or recording of an event. No fiddling is done, normally - and a match with the trees, shrubs, sky outside will be spot on - if calibration of one's set has been done.

This means that morning shows have pinkified faces, lots of orange faces in station promos, quite washed out colour in dramas intending to evoke 'old times'; recent Brit series have pumped up the colour, for some reason - on balance, the colours "feel" like they're correct for the situation, and the known "style" of the producers of the shows.

No fiddling is done? I can gaurantee you that the camera had the white balance set to match the outdoor light which is not true white. If no "fiddling" were done the image would have a blue cast over it.

 

[Wombat]

"Current technology can measure signal fidelity through the reproducing system from recorded input up to the ears with a high level of precision. Comparison between systems electrical/mechanical/acoustic performance can be compared Research has been conducted on the rest of the process, with various aims in mind."




I think you took the chain too far. I think the correct statement would be "Current technology can measure signal fidelity through the reproducing system from recorded input up to the speaker terminals with a high level of precision." Once the speakers and room are in play technically measured "fidelity" of the "recorded input" becomes a mess. You can't really measure the complex 3 dimensional acoustic waveform that you get in the playback room against the electrical signals at the speaker terminals. I'm not saying you can't measure speakers or assess certain aspects of speaker performance in terms of accuracy. But there are aspects of speaker output and room interaction that simply can not be measured in terms of accuracy to the source signal. Audio signals have no such things as radiation patterns. There is no such thing as a speaker radiation pattern that is more or less accurate in comparison to the source signal.

I hold to my statement for measuring the waveform that arrives at the ear, as this is what gets processed by our hearing. If the signal/sound has become a mess beyond the loudspeaker terminals, so be it.

 

[Analog Scott]

I hold to my statement for measuring the waveform that arrives at the ear, as ultimately this is what gets processed by our hearing. If the signal/sound has become a mess beyond the loudspeaker terminals, so be it.

How does one objectively measure the waveform that arrives at one's ear to the source signal with a high level of precision? Can you show me an example of such highly precise measurments?

 

[March Audio]

The issues of bias effects were not really the subject. Bias effects can be controlled if that is what someone wants to do. One can do double blind taste tests with chocolate if they want. I'm betting the results won't change much when it comes to determining if someone likes chocolate or not.

I am a bit skeptical of Sean Olive's conclusions on that subject. What does flat frequency response in a sound system give you with a recording made with microphones that do not have flat frequency responses? When one looks at the vast body of commercial recordings over the past hundred years or so how many were recorded only with mics that had a flat frequency response for 20-20kHz? I'm going to go with pretty close to zero out of god knows how many recordings we are talking about. So what do you actually get? You get an accurate representation of colored microphones and subjectiove choices by recording engineers about microphone placement and subjective choices about final mixes and any post processing that may have been done. Has Olive published the research that proves his claim of transferability? Until I see that I remain skeptical that the body of Olive's tests using one speaker with a very small number of similar "well recorded and relatively neutral" recordings and a regular group of trained listening panel (trained to hear what had already been determined matters) adequately addressed the vast variables of the history of recorded music. Maybe I am wrong and I know he has claimed his results are universally transferable to home audio. But has he shown us how he came to that conclusion? Do we know how many different recordings he used with how many different rooms and different listeners to determine that his results with one speaker, one group of trained listeners and a very small number of similar "well recorded and relatively neutral" recordings were transferable to such a broad range of variables?

Well if we are talking about individuals aesthetic judgements, then clearly it it is very relevant. Peoples aesthetic judgements are very much influenced by external factors other than actual sound quality.

Funnily enough I have been measuring mics very recently (was for the purpose of checking for faults - I spend some time in a studio and doing live sound) and know full well how microphones measure differently and sound different. Interestingly I sometimes see them selected not necessarily by merit, but by personal preference, brand reputation or "everyone uses these for drums" sort of comments.

Again, whatever the subjective decisions of the recording engineer/producer, I dont see the value of having a hifi that is not attempting to be (within reason) neutral. Im not sure precisely what Olive work you are referring to, but the Toole/Olive Harman research was with many speakers over many years and, in my understanding, clearly showed a consistent preference for neutral speakers. Flat smooth anechoic FR and smooth off axis response. Excuse my paraphrasing.

Note the below was not meant to be a calibrated test, just looking for obvious faults, but you can see the mics basic "personalities". If you wanted to get keen about it measure at close distances to see Cardiod proximity responses.

 

[Analog Scott]

Well if we are talking about individuals aesthetic judgements, then clearly it it is very relevant. Peoples aesthetic judgements are very much influenced by external factors other than actual sound quality.

Funnily enough I have been measuring mics very recently (was for the purpose of checking for faults - I spend some time in a studio and doing live sound) and know full well how microphones measure differently and sound different. Interestingly I sometimes see them selected not necessarily by merit, but by personal preference, brand reputation or "everyone uses these for drums" sort of comments.

Again, whatever the subjective decisions of the recording engineer/producer, I dont see the value of having a hifi that is not attempting to be (within reason) neutral. Im not sure precisely what Olive work you are referring to, but the Toole/Olive Harman research was with many speakers over many years and, in my understanding, clearly showed a preference for neutral speakers. Flat smooth anechoic FR and smooth off axis response. Excuse my paraphrasing.

We weren't just talking about individual's aesthetic judgements. We were talking about the validity of those judgements based on the level of knowledge of the individual in regards to the mechanisms of their perception. My assertion was that a person does not need to know the biology and chemestry of taste to know they like chocolate. Bias effects are in play whether one knows those things or not so actually to that specific point bias effects actually are not relevant. One can be an expert on the chemestry of food and still be subject to bias effects when they claim to like chocolate. anyway...

Just because you personally don't see the value of having hifi that is not attempting to be (within reason) neutral. Doesn't mean it doesn't have value to others. what you personally see or don't see is not a universal standard for reality. Ironically enough, you reference Sean Olive's research showing a preference for neutral speakers. Did you know that he has been using a small set of recordings some of which have had added euphonic distortions to the raw mic feeds? What does that say about the value of "neutral?" By what standard is any recording "neutral?" All of his research was done with a small set of recordings. How does Olive or anyone else claim the output is "neutral" when all tests are done with recordings that have built in distortions?

 

[Analog Scott]

Note the below was not meant to be a calibrated test, just looking for obvious faults, but you can see the mics basic "personalities". If you wanted to get keen about it measure at close distances to see Cardiod proximity responses.

View attachment 9510

Yup. Those mics all have a lot of "personality." Are there any in this world used for recording music that don't?

 

[March Audio]

We weren't just talking about individual's aesthetic judgements. We were talking about the validity of those judgements based on the level of knowledge of the individual in regards to the mechanisms of their perception. My assertion was that a person does not need to know the biology and chemestry of taste to know they like chocolate. Bias effects are in play whether one knows those things or not so actually to that specific point bias effects actually are not relevant. One can be an expert on the chestry of food and still be subject to bias effects when they claim to like chocolate. anyway...

Just because you personally don't see the value of having hifi that is not attempting to be (within reason) neutral. Doesn't mean it doesn't have value to others. what you personally see or don't see is not a universal standard for reality. Ironically enough, you reference Sean Olive's research showing a preference for neutral speakers. Did you know that he has been using a small set of recordings some of which have had added euphonic distortions to the raw mic feeds? What does that say about the value of "neutral?" By what standard is any recording "neutral?" All of his research was done with a small set of recordings. How does Olive or anyone else claim the output is "neutral" when all tests are done with recordings that have built in distortions?

1. So, whats your point? People like all sorts of stuff for all sorts of reasons. However the evidence and research shows when tested under controlled conditions they still prefer speakers that conform to a predictable and measurable response characteristic.

Was the Toole research over many years done with very limited recordings? Can you reference that?

 

[RayDunzl]

How does one objectively measure the waveform that arrives at one's ear to the source signal with a high level of precision?

I don't claim a high level of precision, but...

I'll take an RTA of the source and compare to an RTA of the corrected/corrupted in-room sound at the listening position sometimes... generally just examining the peaks...

The mic may not hear what I hear (it has better HF response) but gives me a warm feeling that I'm in the ballpark for frequency response.

---

Old picture below: Some Arvo Part Organ Music playing. Top - room (DSP corrected) RTA. Middle - left source RTA. Bottom - right source RTA.

And for most listening, I'm set to approximate the CD output level with the room SPL, hence the SPL meters on the left.

 

[Analog Scott]

1. So, whats your point?

My point was already made when I originally made the posts you are referencing. Feel free to go back and see for yourself.