Was this aimed at ASR?

[ttimer]

That's not apt. Your speakers shouldn't seek to render everything as a concert hall. They should be made to as accurately as possible recreate the recorded information. If you want the ambience of a concert hall, that's in the recording. And if you want to be able to faithfully reproduce a recording, you need neutral response. If it's in regards to sound pressure level or bass capability, that's a matter of speaker design as well.

A TV can't be A/B tested against a holodeck, but it can be made to faithfully represent a 2D image of a hologram rendered in a movie.

Thats a good point. But the issue of a lack of valid benchmarks against which consumers could compare speakers (and headphones) remains. In order for market forces to push manufacturers towards accurate reproduction, the consumer would need to be able to compare speakers with the "intended sound" and chose the ones which come closest (or read reviews on ASR ).
As an aside, are you sure that TVs or monitors are closer to faithful reproduction or original material than speakers? The "Movie mode" on most TVs looks garish and is more equivalent to a smilyface frequency response than to a linear one. Outside the pro realm (Eizo or NEC), most monitors don't provide faithful reproduction either, unless you calibrate them (the equivalent of EQin for flat response).

 

[bobbooo]

Just out of -- academic -- curiosity: Y'all got a reference for that? Ideally in the peer-reviewed literature?

There you go. From the Clinical Veterinary Science department of the University of Bristol, published in the Journal of British Poultry Science Got to love blind tests!

By the way, to bring it back to audio, a good analogy between sound and food is music is akin to flavours and textures, and hi-fidelity audio reproduction is a recipe to recreate fine dining at home. Follow a chef's (artist's) recipe exactly, and you'll recreate the flavours and textures (music) they intended you to experience. Salt and pepper (DSP) can be added optionally at home to taste e.g. if the chef isn't too great, once the flavours have been tried (music heard), not by piling it in blind during cooking and destroying the balance of flavours (using audio equipment that mucks up tonality or adds huge amounts of distortion to everything). Only blind testing (tasting!) will tell you your actual preference though, with all biases removed. Ok I think that's exhausted that analogy.

 

[Shorty]

Just out of -- academic -- curiosity: Y'all got a reference for that? Ideally in the peer-reviewed literature?

I read it years ago in a non-English (Dutch) newspaper. Google turned up a newspaper article on a study at Bristol University, published in the peer reviewed British Poultry Science Journal (www.dailymail.co.uk/news/article-1032649/Battery-farmed-chickens-beat-free-range-rivals-blind-taste-test-say-scientists.html)
I’ve no idea if this really was an independent study and/or if it was paid for by the poultry industry.

EDIT: bobbooo beat me to it...

 

[MediumRare]

I read it years ago in a non-English (Dutch) newspaper. Google turned up a newspaper article on a study at Bristol University, published in the peer reviewed British Poultry Science Journal (www.dailymail.co.uk/news/article-1032649/Battery-farmed-chickens-beat-free-range-rivals-blind-taste-test-say-scientists.html)

Solid state beats tubes again!

 

[pozz]

How can there be a precise "target curve" when people of different age, gender, country, experience, etc. have different FR preferences?

Just to address this point in case it hasn't been addressed yet.

• Harman research showed that, when doing DBT of speakers, preferences are consistent across age, gender, culture, country and so on. This is an explicit conclusion.
• The target curve is not about speakers, but preferred tone. It was tested by giving listeners EQ controls. This is why Toole says every system should have tone controls.
• For actual speaker design, tone controls are only effective if the speaker has controlled directivity.

My impression is that you have read only a piece of the research, drew somewhat correct conclusions (right in some places, wrong in others) and extrapolated the rest.

The Stereophile article is so vague that if you aren't clear about a lot of background it will seem defendable.

 

[Shorty]

Ok I think that's exhausted that analogy.

Not yet. Expectation bias also plays a role: pour two glasses of orange juice. Add a tasteless purple pigment to one glass. No prizes for guessing which juice people think tastes better.

 

[MediumRare]

Just to address this point in case it hasn't been addressed yet.

• Harman research showed that, when doing DBT of speakers, preferences are consistent across age, gender, culture, country and so on. This is an explicit conclusion.
• The target curve is not about speakers, but preferred tone. It was tested by giving listeners EQ controls. This is why Toole says every system should have tone controls.
• For actual speaker design, tone controls are only effective if the speaker has controlled directivity.

My impression is that you have read only a piece of the research, drew somewhat correct conclusions (right in some places, wrong in others) and extrapolated the rest.

The Stereophile article is so vague that if you aren't clear about a lot of background it will seem defendable.

Pozz, you're not wrong, but not really right either. The two studies need to be taken together - they measure different aspects of the same thing, but with different in-built error-bands. So, hypothetically, a speaker that received below-average scores from a portion of the audience would receive higher scores (by that segment, but not by another segment) if it were EQd - let's say to pull treble down a bit. The manufacturer could achieve the same result via crossover design. It's not necessary that that segment of listeners can be defined demographically, they could be defined biologically (i.e. by hearing perception). Net, there may be two or more segments with divergent preferences that were not discovered in the original data analysis. (The field of organoleptic/taste perception demonstrates this without any doubt.) A subsequent analysis of the same data could reveal new learning. So it's good science to combine the two studies to gain new insights. Sounds like a PhD thesis topic to me!

 

[pozz]

Pozz, you're not wrong, but not really right either. The two studies need to be taken together - they measure different aspects of the same thing, but with different in-built error-bands. So, hypothetically, a speaker that received below-average scores from a portion of the audience would receive higher scores (by that segment, but not by another segment) if it were EQd - let's say to pull treble down a bit. The manufacturer could achieve the same result via crossover design. It's not necessary that that segment of listeners can be defined demographically, they could be defined biologically (i.e. by hearing perception). Net, there may be two or more segments with divergent preferences that were not discovered in the original data analysis. (The field of organoleptic/taste perception demonstrates this without any doubt.) A subsequent analysis of the same data could reveal new learning. So it's good science to combine the two studies to gain new insights. Sounds like a PhD thesis topic to me!

I'm not sure how to reply. Preference isn't really in question; i.e., you will like what you like. The point, as I see it, is how speakers should be assessed.

The most important thing is to give the listener control and ability to exert their preference through their buying decision. This is done in part through their choice of speaker and in part through their ability to control tone manually. But they cannot EQ the sound if the chosen speaker has directivity errors or generally ragged response.

In the Harman research, they controlled for "speaker tone" by having specific studies explicitly use EQ. IIRC they found that individual listeners will try to EQ each tested speaker in an individual way, but will rank them similarly after doing so. This is paralled by the group results: when comparing speaker rankings across different groups of listeners, the results are generally consistent, even if the actual preference scores for each group are not.

There is of course more work to be done but we should be clear about what has already taken place. Most of the criticisms that come up have already been addressed by Harman research.

The error bands, for example, are a measure of consistency, with some listeners being so inconsistent in their rankings and preference scores that their answers can only be random, which means that they are unable to determine by listening alone what they prefer (this is a common audiophile issue: they invent preferences where there are none, as with playback gear with low levels of distortion, since they can't actually tell the gear apart). That's understandable and an element of every subjective study. Others will give the same rankings and preference scores over and over. Combining all of those results will still produce concrete trends, though.

The main problem left to address is calculation of the preference scores using only anechoic speaker measurements (which is another topic).

 

[richard12511]

But music definitely has the edge in triggering subjective emotions imho. It shouldn't be a surprise the debates around its composition, performance and reproduction are always so subjective.

I disagree. Movies do a much better job of triggering emotion than music does, at least for me. I've never had a piece of music that made me openly weep the way a movie like Schindler's List or the Green Mile did. Of course movies have music, but it's only a contributing factor. The story is the most important part, imo.

 

[richard12511]

Part of the problem with audio is that in most genres the "intended sound" could not realistically be reproduced with speakers or headphones. Much less easily compared. You simply canot A/B test a stereo system against a concert hall.

I disagree. The vast majority of music genres don't have a live reference counterpart. Even Jazz often uses loudspeakers. Classical is the only real genre I can think of that usually has a live reference, and classical is a small minority of music.

 

[pozz]

I disagree. Movies do a much better job of triggering emotion than music does, at least for me. I've never had a piece of music that made me openly weep the way a movie like Schindler's List or the Green Mile did. Of course movies have music, but it's only a contributing factor. The story is the most important part, imo.

Ah I definitely live in the opposite world. I find most movies not very interesting, and but given how often my friends or family watch I put a lot of "manual" effort into making them interesting by engaging the content and cinematography (thinking about history, themes, similar directors, use of technique and so on).

 

[b1daly]

The issue is he implies that there are equally valid viewpoints which, in fact, there are not. All there are, are personal, subjective viewpoints that invariably fail the double blind test.

You are eliding two categories of equipment here: the electronic/amplifier vs the speaker/transducer.

There have been many embarrassing failures to distinguish electronic audio equipment in double blind testing.

This is not true for speakers which are easily distinguished in blind testing. This is the entire foundation Tool’s research. Doing the same sort of preference testing comparing converters or amps would not be able to establish a preference ranking because listeners cannot even distinguish between the gear being listened to. (It would result in no meaningful statistical information).

The only basis for saying someone’s subjective opinion is ‘wrong’ or ‘invalid’ would come in the electronic category of gear, because you could prove that if they reported some kind of implied ranking (this sooper-goober box ‘lifted the veil’ from the music) that this was a non-stable preference resulting from subjective illusions.

If a listener makes the same claim about a speaker, the prefer A to B, as long as they can consistently identify such gear in a blind test, then there is no basis for saying their opinion is ‘invalid’.

I tend to not prefer modern speakers with objectively ‘good’ good measurements. There are other speakers I simply enjoy listening to more.

One of Tool’s arguments in favor of ‘neutral’ speakers is that given the range of recordings, you have a better chance of the speaker presenting the audio in a flattering way.

I find my favorite speakers are much more fussy than neutral speakers, and I have to tweak them sometimes based on the specific recording.

I also find that some music sounds just fine on neutral speakers, in particular elevations styles, and just modern recordings in general. But such speakers sound downright ‘wrong’ for a lot of my favorite recordings from the seventies.

There is a simple thought experiment which shows that neutral speakers do not always provide the ‘best’ representation of artistic intent.

The best speakers for representing a recording will be those that were the primary monitors for the production. Since most speakers of the past were ‘inaccurate’ the specific spectral qualities of the monitor get ‘embedded’ in the signal. Given this inevitable truth, in fact accurate speakers will be ‘wrong’. This is a paradoxical quality of studio monitors. A negative ‘echo’ of the speakers deficiencies gets incorporated into the production, and it’s damn hard to get it out.

As a simple example, if you have a studio monitor that has a nice satisfying bass response that is a deviation from flat frequency response, the bass in the mix will be the opposite.

Sound on Sound magazine actually did an article where they searched for the signature of the Yamaha NS10 in mixes and found it (so they claim).

 

[pozz]

I also find that some music sounds just fine on neutral speakers, in particular elevations styles, and just modern recordings in general. But such speakers sound downright ‘wrong’ for a lot of my favorite recordings from the seventies.

This is interesting.

The best speakers for representing a recording will be those that were the primary monitors for the production. Since most speakers of the past were ‘inaccurate’ the specific spectral qualities of the monitor get ‘embedded’ in the signal. Given this inevitable truth, in fact accurate speakers will be ‘wrong’. This is a paradoxical quality of studio monitors. A negative ‘echo’ of the speakers deficiencies gets incorporated into the production, and it’s damn hard to get it out.

As a simple example, if you have a studio monitor that has a nice satisfying bass response that is a deviation from flat frequency response, the bass in the mix will be the opposite.

The Harman research showed that a lot of mixing/mastering engineer EQ decisions were simply undoing the built-in FR of their monitors (and that their mixing rooms were very different from each other in terms of response). So they were trying to go for neutral instead of preserving the monitor/room signatures.

This makes sense since you want the mix to translate, which means that the highest probability for "translation" (sounding good on as many systems as possible) is if there is no specific FR curve in mind for playback.

 

[GelbeMusik]

In my eyes you are projecting feelings and viewpoints to the author that has no support in what was written.

I quoted the parts I were referring to literally. The argumentation was first cheap, too obvious criticism which was given by Toole and Olive themselves (not quoted). Then he accuses people to be so and so, in order to derive counterarguments (offenses?) against that people (!!) (quoted).

( Known scheme in rhetoric )

He lies in saying that the other party lacks self-reflection despite the very fact that us guys exactly does that in using science to back-up the ideas - to the max!

Then he, to go full circle claims for his folks exactly that authority to do so, namely to judge without any objective back-up whatsoever.

This all was written by him with the background of earning money in this field using his "methods".

I won't tell in public what I actually think of him. You may imagine Yourself.

But please for the sake of a civilized discussion here, do not ignore my argumentation and take its content wrong by intent. Thank You so much!

 

[b1daly]

The problem is the conflation of music and audio reproduction. Yes, music can be incredibly emotional, but audio reproduction is not - it's just science and engineering, which should aim to preserve the original emotion and intent of the artist with 100% audible transparency.

This is an oversimplification and unrealistic notion of how the whole system of audio production and consumption works. An audio signal is only perceivable when reproduced by a speaker. It becomes embodied in the speaker, and then excites the air and physical structure of the room. That is the actual ‘medium’.

Producers target the global environment of playback systems they expect the work might be listened to on, as best they can.

So the ‘message’ is not truly independent from the ‘medium’.

Tool refers to this concept with his idea of the ‘circle of confusion’. I’m not persuaded with his analysis of the problem though, because I think it is overly idealistic and unnecessary. (Part of his solution is there being more objective standards for recording studio monitoring, kind of getting a common reference between production and playback systems). The world is not going this way, and the economic infrastructure that supports the production and consumption of music and audio is too complicated and diffuse. The cost of implementing more standardized audio is not worth the benefit.

Real-world playback systems are always changing, varied, and the vast majority of people listen on systems that are far from the types of theoretical concerns we are discussing here.

 

[richard12511]

Ah I definitely live in the opposite world. I find most movies not very interesting, and but given how often my friends or family watch I put a lot of "manual" effort into making them interesting by engaging the content and cinematography (thinking about history, themes, similar directors, use of technique and so on).

I may have (or at least come close) to shedding a tear a few times with music, but I've never had that several minutes of uncontrollable sobbing experience I've had with a few movies.

 

[pozz]

I suspect it comes from the fact that speakers (audio in general?) is so simple to produce that any moron with space on the kitchen bench can assemble and produce finished products that works without prior knowledge. The audio industry is riddled with small producers without tools, resources or technical understanding - and the reviewers praise this stuff.

If technical brilliance is to be expected, we need big companies with resources and rigid enough standards that some Absolute-moron cannot possibly achieve it from his kitchen table. Genelec, Neumann, Harman/Samsung, Monitor Audio, Focal, Canton, Kef etc are examples of companies that can produce products of high enough standards at acceptable prices if they choose to.
In my mind it would be better for us audiophiles if we forced them to.

Hopefully that would rid the world of one-man shit-shows with creative solutions to problems they don't understand. We avoid cars that doesn't score 5 stars at EuroNCAP crash testing, I would assume we would do the same if standards were applied to audio as well - at least in the more expensive part of the spectrum.

At the same time, speakers are a bunch of opposing compromises that can never be perfect because of it. Let's say you can have broad dispersion or perfect directivity, what would you choose? Or perfect spinorama, but no bass? Or the same but with tons of distortion?
In TV's you know that you can measure whatever goes in comes out, but when it comes to speakers that will never happen. You would need standards for different criteria, perhaps like the labels on a tire?
You know, noise, wet grip, dry grip, rolling resistance etc.

In my eyes you are projecting feelings and viewpoints to the author that has no support in what was written. The author didn't show a misunderstanding of the research in the article, he pointed out quite specifically that the research intended to find averages and avoided stand-outs like hearing-impaired people to make the process simpler.
That's true, of course.

While I don't agree with the author philosophically, I can understand his position and why he assumes it. There doesn't need to be a lack of understanding or contempt involved to disagree.

Just to add on to what you said. TVs were developed using much, much more psychovisual research.

The same is true now with AR/VR/XR/MR. The psychovisual aspect is much more developed and there are many more tools and apps using it specifically (for example, removing occlusions or creating accurate dimensional projections in space).

Psychoacoustic research, on the other hand, has tended to be medical (diagnosing hearing loss specific to certain illnesses or injuries) or optimization-based (e.g., bandwidth requirements for speech intelligibility).

 

[CDMC]

If departure from flat frequency response and deliberate resonant cabinet is goal of the designer, it should be mentioned in the respective speaker advertisement/s. I may be wrong but it seems article is to justify high end bad measuring speakers.

Regards

I agree. Take Nelson Pass, he readily admits that he voices his amplifiers with low order harmonic distortion, as it is pleasing. From his website:

"Many audiophiles believe that 2nd harmonic is to be preferred over 3rd harmonic. Certainly it is simpler in character, and it is well agreed that orders higher than third are more audible and less musical. However when given a choice between the sound of an amplifier whose characteristic is dominantly 2nd harmonic versus 3rd harmonic, a good percentage of listeners choose the 3rd.

I have built many examples of simple 2nd and 3rd harmonic “types” of amplifiers over the last 35 years. When I say “types” I mean that they used simple Class A circuits described as “single-ended” versus “push-pull” and so tended to have a 2nd harmonic versus 3rd harmonic in the character of their distortion, but were not made to deliberately distort.

Anecdotally, it appears that preferences break out roughly into a third of customers liking 2nd harmonic types, a third liking 3rd harmonic, and the remainder liking neither or both. Customers have also been known to change their mind over a period of time."

https://www.passlabs.com/technical_article/audio-distortion-and-feedback/

 

[Thomas savage]

This is Stereophile trying desperately to be part of the conversation and stay relevant.

Print mags are out and stereophile fell asleep at the wheel while all the audio forums were establishing themselves.

They are , as my dad would say after becoming exasperated by some non compliant bit of something he was trying to fix ..., fu... What best forum'd

 

[SIY]

I agree. Take Nelson Pass, he readily admits that he voices his amplifiers with low order harmonic distortion, as it is pleasing. From his website:

"Many audiophiles believe that 2nd harmonic is to be preferred over 3rd harmonic. Certainly it is simpler in character, and it is well agreed that orders higher than third are more audible and less musical. However when given a choice between the sound of an amplifier whose characteristic is dominantly 2nd harmonic versus 3rd harmonic, a good percentage of listeners choose the 3rd.

I have built many examples of simple 2nd and 3rd harmonic “types” of amplifiers over the last 35 years. When I say “types” I mean that they used simple Class A circuits described as “single-ended” versus “push-pull” and so tended to have a 2nd harmonic versus 3rd harmonic in the character of their distortion, but were not made to deliberately distort.

Anecdotally, it appears that preferences break out roughly into a third of customers liking 2nd harmonic types, a third liking 3rd harmonic, and the remainder liking neither or both. Customers have also been known to change their mind over a period of time."

https://www.passlabs.com/technical_article/audio-distortion-and-feedback/

Still waiting for him to demonstrate that the “voicing” via nonlinearity is even audible, much less preferred.