[pinnahertz]

When you say DSP 'can in fact neutralize "analog colorations"', you do realize that means it is adjusting frequencies (IE colorizing)??

Adjusting frequency response is one possible DSP function, but when working in the time domain, much more is accomplished. Every single speaker or headphone (analog systems) have serious problems both in frequency and time domains. Those problems result in what you're terming "colorations". I have no issue with the use of the term. You may like those colorations, or not, or just not care. The environment a speaker is in also has similar impact, and results may or may not be desirable. With proper measurement and convolution it is possible to compensate for, to a very large extent, many of those deficiencies and end up with a system that is more neutral, faithful, high-fidelity...you pick your favorite term...than before. You can also match a target curve for a particular application, which is most often done. The end result, when proper measurement and targets are applied, is not more "coloration" but rather a reduction in the flaws and deficiencies that caused coloration in the first place. It's been established that people prefer smooth and neutral response. DSP is a way to get there.

You might want to assume the latest digital processing is the easiest and best in sound, but I think it's just another cycle.

It's always another cycle. Todays DSP products and solutions are the best yet, but there's always higher speed, more resolution, more memory, etc. I feel most of the work that needs to be done is in the area of collecting complete and representative measurements. It's impossible to apply full DSP correction unless you know what you're correcting for. In my work I've found measurement is actually the key to excellent sounding DSP equalization, and unfortunately, it's not as intuitive or as easy as one might hope. Can't tell you how many hours I've spent only to trash the results and start over using different measurement techniques. The fully automated solutions are only approximate. I'm always hand-tuning and re-measuring anyway. Hence my feeling that measurement improvement is the next step. I think DSP capability today is more than adequate to accomplish most of what we need to do.

 

[tipsyboo]

This was the best read ever.

 

[bigshot]

There are a lot of different types of DSPs intended for all kinds of purposes. I don't think you can make any generalization about what DSPs do or don't do, you can only say that they do something. They can create coloration or uncolor. They can create time shifts or eliminate them. They can simulate unnatural ambiences or make the room ambience sound more natural. I use DSPs all the time in my speaker system. I wish there were sophisticated DSPs that I could easily use with headphones too. I believe that DSPs are the future of high end audio.

 

[old tech]

Thanks for the detailed post. Some questions:

- Has anyone ever measure the upper frequency limit at which the ear is no longer generating auditory nerve signals?

- Has anyone ever measured response of auditory regions in the brain for high-frequency test tones, to see the upper limit of frequencies?

- What blind listening tests have been done to compare music with and without 'ultrasonic' content? What are the results of those tests?

I can answer the last question. Commencing around 20 years ago, I made probably around 500 CDs from LPs in my collection. Firstly they were done in a well known Australian studio and then later, at home when consumer software/hardware became available. As I put a lot of effort into this and a large number of LPs, I regularly conducted blind tests against the donor turntable on the same stereo. Apart from some faulty transfers along the way, netiher I nor friends who participated in the tests could pick the CD apart from the LP (now remember CD's top out at around 20khz, while LPs have noise which goes higher than that). Neither could the professionals at the studio tell them apart and that was 20 years ago.

A better example is the M&M 2007 year long study comparing SACDs and DVD-A clean and downsampled to 16/44. The subjects of that study included trained musicians, record producers and audiophiles, including allowing the subjects to use their own music material, on their own stereos, in their own home and in their own time. Again any ultrasonic content would have been removed when downsampled to 16/44 and no evidence was found that the listeners could pick the difference.

Apart from those examples, what about the 1000s of people who load up a hi res track in foobar and then do a proper DBT and cannot tell the difference?

How much evidence do you need? Surely the onus of proof is the other way round.

 

[old tech]

I drive cars on race tracks, I know that topic well. The analogy isn't good because 'going faster' is precisely measurable using the simple unit of time, and the physics of what contributes to going faster is very well understood and also precisely measurable. We don't have anywhere near that level of understanding and measurability when it comes to auditory perception.

Actually it is quite a good analogy. Some cars are deceptively fast and some are deceptively slow. There are many instances where people might ride in a slower accelerating car and be convinced it is quicker than the faster car, until they see the precise measure - like audio in a way. Some are convinced that CD is not completely transparent, despite the precise measurements while others say vinyl has superior fidelity despite not even needing precise measurements to demonstrate it is not.

Like audio, beliefs and preferences can affect how an individual responds to a car, regardless of measurement. I know quite a few people that prefer 1960s muscle cars over modern cars. They would very much prefer driving a 60s Mustang or Camaro over say a current model Kia Stinger, even though the Stinger is quicker, handles much better, has better brakes, more precise steering, more comfort and indeed, measures objectively far better on any relevant measure.

 

[castleofargh]

OK, spike in frequency isn't ringing to you....got it.



No, treble attenuation or compression is not "default", but I said often used. I have not even gotten into what my experience with DSP has been, and no, not all my applications where what my generality was (hint: you do realize Bose is one of the most popular audio companies). I don't care what DSP was like in the 70s (before CD): I'm at least talking about popular 16 bit audio of the late 80s. At least DSP processing of 48khz+ should be considered.

When you say DSP 'can in fact neutralize "analog colorations"', you do realize that means it is adjusting frequencies (IE colorizing)?? You might want to assume the latest digital processing is the easiest and best in sound, but I think it's just another cycle.

about ringing, @pinnahertz was just explaining that in this context, it doesn't refer to a subjective impact(my ears are ringing), but to a characteristic of the signal(pre or post ringing depending on the type of filter used for the EQ). the biggest effects can be observed with impulse responses and a strong low pass filter. that's the type of ringing @prescient was referring to in his post.

about coloration, alteration, degradation of the sound, it's really a matter of reference. if I could purchase flat speakers, put them in a room, sit in a chair, and get clean flat sound, then room correction through DSP would be a mistake. because all I would do would be "coloration".
but as reality is more likely to have us end up with sound at the listener's position being a mess of non flat transducers, room reflections, comb effects and what not. proper settings in a DSP can effectively result in higher fidelity and a signal closer to neutral(if that's what we were aiming for).

 

[gregorio]

I saw a recent commercial for a migraine remedy "for people who have 10 - 14 migraine days a month".
The manufacturer was very pleased to announce that their drug reduced headache days by 65% - while the placebo only reduced them by 40%.
My reaction was along the lines of:
[2] "Gee, we can reduce migraines by 40% just by thinking the right thoughts... there should be something we can do with that."
[2a] "Wouldn't it be nice if we could just teach people to get the placebo effect.... maybe by meditation or some such method?"

That's effectively a common audiophile response. By that I mean I've often seen it used as a response, typically something along the lines of: "If it [some bit of kit or format, etc.] sounds better, what difference does it make whether it sounds better because of placebo or some real affect? The end result is exactly the same, it sounds better and therefore I've wisely spent my money on something that sounds better!". - The problem is of course that placebo effect is unpredictable and volatile. It may work for one person and not another and even if it does work for someone on one occasion, it may or may not work on the next occasion.

2. There IS something "we can do with that" in the audiophile world and virtually without exception something is done with it!
2a. We can and routinely do "teach people to get the placebo effect" not with meditation though, the "some such method" is called marketing!!

Placebo is caused by the interaction of various biases and marketing exists to manipulate biases. Therefore, even if placebo works for someone consistently, it's likely that some form of marketing will come along which changes those biases, reduces or changes that placebo effect and creates a new placebo effect. In fact, the audiophile world almost entirely exists solely on this principle!

Let's assume that I have the opportunity to purchase a CD for $15 or a high-res file version of the same content for $20.
Let's assume that it's a given that they are identical except for the fact that the high-res version has some extra "ultrasonic" content.
Let's also assume that I have actually had an opportunity to listen to both casually and been unsure whether I noticed a real difference or not.
[1] I PERSONALLY would still feel mildly uneasy purchasing the version that I know is technically less accurate....Call it "a negative placebo effect" from knowing that it is NOT the most accurate copy available.
[1a] I would wonder, at least somewhat, whether, after buying new headphones next week, I might discover that an audible difference did in fact exist. (I'm not buying an improvement; I'm buying insurance against a risk.)
[2] And, yes, this is based to a degree on personal experience.
A very long time ago I collected a significant amount of music in MP3 format. Then, after discovering that I could hear a difference in many cases ...

1. As mentioned, that's called a "nocebo effect". In your example, a nocebo effect based on the fallacious belief that the CD is somehow less accurate.
1a. You're buying an insurance against the risk that you may not be a human being?

2. How is that based on your personal experience? The MP3 format removes masses of information from within the human hearing range, to hear a difference does not require you to be some species other than a human being, it just requires you to have acute human hearing/listening (perception) skills. The difference between MP3 and CD does not equate to the difference between CD and so called hi-res.

Your posts indicate you don't understand what "inaudible" means or the difference between "imperceivable" and "inaudible".

[1] ]However, with all the knowledge we have about cars, and speed, and aerodynamics, you would think that it would be a trivial assumption that the company with the most money to spend could reliably design the fastest car.
[1a] Or could it be that even a simple subject like that is complex enough that we CAN'T work out all the variables carefully enough to predict the outcome reliably from the specs?
[2] Could it be that "a simple audio signal" isn't quite as simple as some of us would apparently like to believe either?

1. No, I would not think that, I can see how some might though.
1a. Even if you had all the specs you still couldn't predict the outcome. Just with the aerodynamics alone we have a chaotic system, unless the races were only to occur within a specific wind tunnel, and that is just one of numerous variables, some others of which are also chaotic systems. A chaotic system is by definition unpredictable or at best, only predictable to within a probability (of less than 100%).

2. No! Audio signals, regardless of their complexity, are NOT a chaotic system. They are entirely definable/predictable and have been proven to be so. To state or prove otherwise would require disproving the already well established and proven mathematics of Fourier, Shannon and a bunch of others.

G

 

[gregorio]

[1] The facts are what folks like you and Shepard want to conceal with statements like "ignore the numbers" and "use your ears".
[2] I was able to create 'after' waveforms similar to those from 'remasters' by using, gues what - compressors and limiters! The same bloated waveform signature. Those are the FACTs.

1. Either it's all some massive conspiracy of engineers against audiophiles or it's simply you deliberately misquoting "statements" in order to justify your agenda.
2. Those are the irrelevant and out of context facts, which you've misunderstood. What's surprising is that you continue to misunderstand despite it having been explained and demonstrated to you!

I'm not going to go through it all again with you for about the fourth time. I'll leave it to others to work out for themselves if this a conspiracy and the real "facts" or you just misrepresenting the facts to further your own agenda!

I drive cars on race tracks, I know that topic well. The analogy isn't good because 'going faster' is precisely measurable using the simple unit of time, and the physics of what contributes to going faster is very well understood and also precisely measurable. We don't have anywhere near that level of understanding and measurability when it comes to auditory perception.

What has a level of understanding and measurability of auditory perception go to do with it? We're talking about the limits of audibility, NOT the limits of perception!

G

 

[prescient]

pinnahertz thanks for the replies. With respect to the ringing the reason I asked is that I had seen people on message boards indicate it was a potential issue, but I've never noticed it as I suspect my room is a much larger issue. I am interested in understanding the potential downsides of the use of DSP despite the fact that they are probably minimal in the grand scheme of things.

I actually have Dirac, a minidsp DDRC-88a, and minidsp headphone measurement rig to play around with. The DDRC-88a is for the living room, band but I'm planning to measure my headphones and play around with creating EQ curves to see how that works out. I also have this plan to build active speakers with DSP built in at some point in the future just to see if I can do it, but obviously have no clue how to do it at the moment.

 

[Phronesis]

Probably not.
But we might find out that the air drag on the rear view mirrors really is enough to make a difference.
And Olympic swimmers now wear swimsuits made of special fabric - designed to slip through the water a tiny bit more easily.

However, with all the knowledge we have about cars, and speed, and aerodynamics, you would think that it would be a trivial assumption that the company with the most money to spend could reliably design the fastest car.

In fact, I would wonder why we bother to have races at all....
Can't we just analyze the cars thoroughly and calculate which car will win - based on the specs?
Isn't it obvious which one can deliver the most torque per weight - and so get around the track the fastest?
Yet, oddly, we still need to build prototypes... and actually test them.

Or could it be that even a simple subject like that is complex enough that we CAN'T work out all the variables carefully enough to predict the outcome reliably from the specs?

Could it be that "a simple audio signal" isn't quite as simple as some of us would apparently like to believe either?

Yes, this is generally true. The biggest budgets for developing race cars are in Formula 1. There are literally hundreds of people in each team to support development of two cars for two drivers. The knowledge is there for engineers to develop cars that are nearly optimized without too much difficulty, with the result that the difference in lap time between the fastest and slowest cars is about 1-2% in a field of 10 teams. But over the course of dozens of laps, that small difference in lap time amounts to a difference between having a shot at winning vs being near last place. Squeezing that last bit of performance out of the car is difficult because of the complexities, design tradeoffs, and small uncertainties which are involved.

With audio, it may similarly be the case that it's not difficult to design very good gear, but squeezing out the last bit of 'sound quality' is challenging, with the further challenge that we can't necessarily be confident that we can measure 'sound quality' in a purely objective way.

 

[Phronesis]

If you had the two samples lined up with each other and looping, you could have put them through a switch box and directly compared them with no time between samples. The problem you had was related to auditory memory, and that has been extensively tested.

The same sound is played back over and over again by the app. So I think we can assume that the sound is objectively the same in each case, and what's changing is my perception of the sound (which, yes, involves auditory memory).

 

[Phronesis]

Actually it is quite a good analogy. Some cars are deceptively fast and some are deceptively slow. There are many instances where people might ride in a slower accelerating car and be convinced it is quicker than the faster car, until they see the precise measure - like audio in a way. Some are convinced that CD is not completely transparent, despite the precise measurements while others say vinyl has superior fidelity despite not even needing precise measurements to demonstrate it is not.

Like audio, beliefs and preferences can affect how an individual responds to a car, regardless of measurement. I know quite a few people that prefer 1960s muscle cars over modern cars. They would very much prefer driving a 60s Mustang or Camaro over say a current model Kia Stinger, even though the Stinger is quicker, handles much better, has better brakes, more precise steering, more comfort and indeed, measures objectively far better on any relevant measure.

Yes, agreed. For example, the sound of a car can certainly influence the perception of speed. For cars used in a non-competitive setting like road use, what really matters isn't how fast the car is, but rather the how enjoyable the experience of driving it is. An objectively less capable car can be more fun, and often is for a lot of people.

 

[Phronesis]

That's effectively a common audiophile response. By that I mean I've often seen it used as a response, typically something along the lines of: "If it [some bit of kit or format, etc.] sounds better, what difference does it make whether it sounds better because of placebo or some real affect? The end result is exactly the same, it sounds better and therefore I've wisely spent my money on something that sounds better!". - The problem is of course that placebo effect is unpredictable and volatile. It may work for one person and not another and even if it does work for someone on one occasion, it may or may not work on the next occasion.

I certainly agree that the first priority needs to be on making sound objectively better, and that it wouldn't be prudent to rely on placebo effects or be duped into spending a lot of money for them. And if testing is done to show that something is objectively better (or just as good as something which costs a lot more), that knowledge itself can be a source of a useful placebo effect.

A recent example is that I compared Focal Utopia (about $4K) and Clear (about $1.5K) headphones. I went into the comparison assuming that the Utopia would sound better, and the question is whether they would sound better enough to justify the much higher cost. While I couldn't do blind testing, I did a lot of short- and long-term listening tests in various ways, using controls in various ways and sometimes keeping switching time under 2 secs, and came to the conclusion that I prefer the sound of the Clear, so that's what I bought. Not only did I wind up saving money, but I now have a sort of placebo effect from having confidence that the Clear sounds better than the Utopia (to me), and I've avoided the negative placebo effect that would have come with having bought the Clear to save money without sufficient comparison to the Utopia, thereby leaving me wondering what I was missing out on by not having bought the Utopia.

 

[old tech]

1. As mentioned, that's called a "nocebo effect". In your example, a nocebo effect based on the fallacious belief that the CD is somehow less accurate.
G

What many audiophiles don't realise is that the nocebo effect can actually reduce your enjoyment of music.

As an example, for several years many audiophiles believed (and still believe) that the original 1984 Beatles Abbey Road CD is the digital holy grail and the 1987 version barely listenable. They use terms to describe the 1987 release as bland and flat, nowhere near as good as the original LP, while the 1984 version is described as 'analog sounding', sounds just like the original LP and so on. I had both, and never really preferred one over the other until I started reading those threads and then noticing how much better the 1984 CD was and even worse, the 1987 CD became almost unlistenable.

Then there was this long thread on the Hoffman forum where again people were bagging the 1987 CD. One of their more grounded and rational members argued that the difference between the two CDs is very minor and really only noticeable when concentrating heavily between the two (his opinion was that the 1987 CD was probably one generation tape older, but still good quality). He was laughed off for a while so he set up a double blind test. Not many could pick which was which - I and some others managed to pick out the 1984 as having a touch more clarity in certain areas, but it was so minor - so then the usual arguments abounded about whether he set up the test correctly or honestly - but it was all verified.

Anyway, my point about the nacebo is that once I knew that there was virtually no difference between the two, the 1987 CD sounds great again. I was previously hearing it as sounding relatively bad, simply because of an audiophile mastering myth.

 

[Phronesis]

Head-fi actually struggles with both placebo and nocebo issues. Reports based on placebo effects make some products 'hot' and cause people to waste money, while reading a thread where someone reports that your beloved product doesn't sound any better or doesn't sound good can adversely impact people's enjoyment of their beloved products.