Does Tonal Balance Affect Perceived Pace and Perceived Resolution?
- Thread starter Ron Resnick
- Start date
They don't think designing a speaker that is correct in phase and time, meaning it makes a right triangle in the step response test, and focus on frequency response on and off axis, resonance control, bass loading etc. but not time-alignment. It is also quite hard to do and requires a lot from the drivers, as you nearly always need to have a 1st order crossover to achieve it. A 4th order L/R is in phase but it has group delay that makes it basically impossible to physically time align.
One thing I had in speaker design is when the midrange is in inverted phase on a three-way design...I can hear that this driver is pushing inward when the other two drivers push out...even though the frequency response sums flat.
Does this impact PRaT? I think it does and the reason is that I have heard some speakers before and after DSP time-alignment and the sense of PRaT was very noticeable. I am talking only about time-alignment and not room correction that includes frequency response adjustment. Once was a demo of a B&W 802 where they measured the speaker's time response and generated an inverse function on a digital recording. The 802 had at that time quite high order filters (even 6th order I think for some drivers). The improvement in the believeability of the speaker was nothing short of phenomenal. Of course it still had all the issues with the driver's self noise, including this horrible woven kevlar midrange the flexes like mad and is practically a bending wave driver but the improvement in timing and pace was dramatic.
Believe what you want but I have heard what time coherence does both before and after it is applied and it matters.
A lot of active speakers with DSP are now time alignable as digital delay can be applied based on the listening distance.
Quite a lot of people use something like Dirac live to time align their system.
Lots of people think using DSP correction sounds better and I'm sure it does with some speakers.
There’s over a century of research in whether human hearing is sensitive to phase going back to the 1840s by the German physicists Ohm and Helmholtz (both of whom concluded we are not). This “tone deaf” viewpoint has shaped many decades of research into DSP. Recent work has shown that we seem sensitive to certain types of phase information (e.g., like the sound of applause). Our hearing evolved over millennia to help us detect predators and it is likely that some phase information may have been useful. But human hearing is hugely complex. At its most sensitive point the eardrum moves less than the width of a hydrogen atom in detecting the faintest sound we can hear! You have to wonder how the brain can figure out signal from noise under such conditions when air molecules are probably randomly perturbing the eardrum by much larger amounts.
There are basically 2 camps.
1) Where FR is the most important
2) Where time and phase of the time domain waveform is most important.
-For tones and woodwinds, french horns, organs, etc. there are no strong impulses happening.
- And for percussion it is mostly impulse.
- Then we have some instruments, like say a piano, where there is a hammer and a percussion (sharp attack) followed by some more steady state like tone, which decays.
Ideally we have both a good FR and a good time domain response.
Without all that people mostly fix the time domain with something like DIRAC live, on a speaker with a good FR but poor TD response.
Or…
They fix the FR with a speaker with a good time domain response.
Or…
They just live with the speaker with no correction.
The consensus seems to be that PRaT seems to be positively affected by DIRAC.
It is less clear whether PRaT is affected with EQ.
Obviously I am in Camp #2 with @sbnx , @godofwealth , @morricab , @Robh3606 etc.
In the mean time we have a lot of side discussions on power, ICs, and speaker cable cables and gawd knows what else… trying to bring back what is lost at the very end of the playback chain at the speaker.
While tho etchings may have some impact, they are not fixing an engineer design decision that discounts what @DaveC described as:
There can only be one correct answer as to what the ideal time domain response plot should look like.
Why do most step function look different?
1) Where FR is the most important
2) Where time and phase of the time domain waveform is most important.
-For tones and woodwinds, french horns, organs, etc. there are no strong impulses happening.
- And for percussion it is mostly impulse.
- Then we have some instruments, like say a piano, where there is a hammer and a percussion (sharp attack) followed by some more steady state like tone, which decays.
Ideally we have both a good FR and a good time domain response.
Without all that people mostly fix the time domain with something like DIRAC live, on a speaker with a good FR but poor TD response.
Or…
They fix the FR with a speaker with a good time domain response.
Or…
They just live with the speaker with no correction.
The consensus seems to be that PRaT seems to be positively affected by DIRAC.
It is less clear whether PRaT is affected with EQ.
Obviously I am in Camp #2 with @sbnx , @godofwealth , @morricab , @Robh3606 etc.
In the mean time we have a lot of side discussions on power, ICs, and speaker cable cables and gawd knows what else… trying to bring back what is lost at the very end of the playback chain at the speaker.
While tho etchings may have some impact, they are not fixing an engineer design decision that discounts what @DaveC described as:
There can only be one correct answer as to what the ideal time domain response plot should look like.
Why do most step function look different?
More like obsessive compulsive disorder.
Rob
You don't suffer from this?? What's wrong with you?
it's relatively easy to handle bass issues such as deeps with sub (there are other issues that sub can help) that you cannot fix with only 2 speakers.
It's not so clear cut. Some aspects of time/phase are audible and some are not, with rare exceptions that would be examples of poor design or an obvious mistake, it has nothing to do with the perception of pace and timing. Not that poor design is rare, I have found every single back-loaded horn I've heard have bass timing issues, and the one exception is arguably not really a BLH. Almost everyone here has heard a ported or bandpass design with out of time bass at some point. OTOH, I've heard absolutely great ported boxes from Focal, TAD and ELAC's designs from Andrew Jones, etc.
I guess I'm in the camp with almost every single speaker designer on the planet, you know...professionals that do this for a living. But hey, everyone is entitled to their own opinions and beliefs regardless of whether they align with reality. Speakers are an exercise in choosing compromise based on what the designer believes is most important within the constraints of cost and size. If you choose to prioritize things that don't really matter you make the design worse than it could be otherwise.
Some engineers have latitude to make the compromises others are part of a larger business team.
It is like saying that “We hear what the artist intended,” which is generally BS. And we hear what the entire chain of recording and mastering intended.
So we take all these compromised designs (or designs of compromise) and get a tapas menu of speakers.
I think ^That^ is a called an argument from authority?
There are speaker designers and there are engineers.
The reality is that some speakers are designed in a way that makes their engineered outcome comport with a more correct acoustic performance, and some are compromised to make parts cost lower, manufacturing cost lower, or other things easier… or meet size goals etc.
They might sound OK, they might sound good, they may look impressive, and they may meet WAF requirements. But they still might be compromised from correct acoustic performance in favour of some other compromises.
It is just a fact that all the designers come up with a broad set of outcomes, and some of those actually are not in the “almost every single speaker designer on the planet” group… and their speakers are more correct in acoustic performance.
IMO, It is certainly safer to start a system with things that are more correct, and speakers have the lion’s share of effect.
In any case, as I mentioned before, I am in the smaller group… the larger group is by definition “the majority”.
That is fine with me. I might be mistaken.
^yep^ It is certainly complicated.
Whether the more correct acoustic performance delivers PRaT seems to be somewhat correlated.
You might be right.
And Toole, etc and ASR rate frequency response very highly.
It is just that the majority of my circle have similar speakers.
Either planners, Dunlavy, Spika, Thiel Vandersteen. So it seems odd that we mostly have fallen into the time/phase correct group.
As an engineer are you saying that step response, transient response and impulse response are less important in your experience?
I think these things all have their audible limits. I was really hesitant to answer morricab because this is always how discussions on "time alignment" go. It's even more predictable than a discussion on the audibility of cables. I don't discount what people hear, but sometimes I am skeptical of what they think is the cause of what they hear. For example, blaming time domain error on hearing the cymbals off pace with a midrange instrument is, outside of something that's really poorly designed or broke, is absolutely not possible. Even with a relatively worst-case 360 degree phase shift the times are too low. I think it's far more likely this delay is on the recording, humans aren't perfect. Could also be room acoustics. That was actually one of the problems with drum machines when they were a new thing, they were too precise and it sounded unnatural. So error was introduced to the timing of drum machines to make them sound less precise, less mechanical and more human. A well designed speaker's error in time domain is simply not an issue, especially at the listening position.
Even a speaker with 1st order crossovers are not perfectly "time aligned" everywhere. This type of speaker is simply easier to get right because it's simpler. Many years ago, and even in some cases today, the simplicity led to a better outcome, and in those times single drivers were much more popular than they are today too, because poorly designed multi-way speakers were more common than not. If you hear a 20 year old YG or Wilson speaker compared to what they are today, there is no comparison, the new speakers are massively superior and in a large part it's due to the crossover design. I went to a YG open house and learned they hired an engineer specifically to help with crossover design and the result is night and day vs previous speakers.
The main issue with simplicity in a speaker is the compromise in reproduction of complex music. If this isn't a priority to the listener, then a relatively complex 3+ way speaker may offer the wrong set of compromises, especially if cost is an issue. With an unlimited budget you can have a system that does everything well, but for most, we must compromise on cost and/or size of the speaker.
It is funny though, after all this my own speaker, the one I designed, built and will one day offer for sale, is an augmented point-source horn where the midrange horn covers from 400-12,000 Hz and has fairly ideal time and dispersion measurements within it's operating range. I've tried different crossovers and there are some advantages to not even using one at all, but for a lot of music and higher SPLs, a 2nd order bandpass is the best compromise with a 4th order lowpass for the woofer and matching 2nd order for the tweeter.
My speaker excels at PRaT and is exceptionally good for low-volume listening too. But the reason for this isn't it's time domain behavior, it's the overall design of the midrange driver, which is custom made for this application, and the horn improving efficiency and dispersion pattern helps too, as does the associated SET amp I also designed and built to be an ideal match for that particular driver.
Also, as much as I respect Andrew Jones, his coax designs with the goal of improved "time alignment", causes more issues than it solves imo. The TAD driver with the Mg mid and Be tweeter does some things exceptionally well, but what those kind of drivers can't do is that last little bit of the speaker disappearing act. I think the midrange cone acting as a waveguide for the tweeter is the main issue, but it could be something else. I'm also not alone in this observation, it's in a Stereophile review too. While the Be tweeter is one of the best in the world, I think other speakers are overall more transparent and AMT and ribbon tweeters can be even better. I also dislike the TAD 7" carbon woofers, way too much euphonic distortion. The 15" Acoustic Elegance woofers I use in my horn speaker are massively superior.
That sort of implies that they did not have an competent engineer before?
Or that maybe some state of the art in XOs changed.
At 400 Hz, a 1 cycle delay is around 3/4 of meter, or 2.5 milliseconds
And 2 cycles 1.5 meters…
It seems like a lot of either distance or time.
I thought that the “near window” of time was somewhere under 5 msec ??
Are there step function plots for your speaker?
If your speaker has PRaT, and the step function, or impulse response, deviates from ideal… then I suppose we can put to rest whether good time domain performance is required for PRaT.
However it makes it difficult for others to have anything in terms of solid engineering metrics to base what they get, if the metrics do not apply to what is heard.
I know that most people have an amazing ability to for having noise dominated sound, and still being able to fill in the blanks especially with lip reading cues.
Maybe a lot of people also have the ability to take time smeared sounds, and put it back together as it should be.
If nothing else, the topic is at least interesting.
It depends on how you look at it. Things are better today... speakers, cars, computers and phones. Did we not have competent engineers in the past? I think we did, but we make progress over time.
I use DSP for the bass. In any case, this is too far OT. It's just one example. It's also a great example of why many audio manufacturers struggle with how much info, specs and charts to make public. People make their mind up based on a mistaken idea of how the info given correlates to what they hear. For example, the idea interconnect cables need to have the lowest capacitance possible. The problem with that is it's trivially easy to make an ic cable with almost no capacitance at all, but it'll have no noise rejection either. One of the most common ic cable geometries around, a star quad, is higher capacitance vs a twisted pair yet has better noise rejection and is often chosen over a twisted pair as a result, and in the case of mic cables may be 50m long. And audiophiles will worry about a 1-2m long cable and choose it on the basis of a capacitance spec.
With time, sensitivity varies a lot depending on the frequency and resulting behavior. We can easily hear cancellation in the crossover range withing a certain range of frequencies, but outside that range we simply don't. This is why it's not a requirement to have only coaxial speakers and make phase vs frequency behavior the most important design criteria. Measurements of comb filtering at high frequencies seems alarming, but thankfully we don't hear the same way the mic records a signal. I'd argue one solution to reduce this, a coax driver, creates more problems than it solves because the design priorities are not what I'd personally choose, and using the midrange's cone as a waveguide is worse than using a separate tweeter. YMMV.... some claim to hear the smallest minutiae in time domain behavior that really isn't in the realm of possibility, and attribute that to many different things. Audio is a complicated system that teaches us logic and some degree of humility if we're willing to pay attention. It's similar to diagnosing problems with cars, an experienced mechanic isn't often going to tell you what's wrong based on a simple description without investigation, yet we hear something in an audio system and are quick to assign a cause to what we hear without enough information.
Attachments
This shows the audibility of Group Delay if that's what you are asking?
Rob
No - I am talking about that that we hear a slap echo as different from the direct sound, when it is some milliseconds delayed.
Too close in time, then we cannot hear it as separate.
Tonal balance can be a frequency response error.
But tonal balance can be changed by distortion too.
To your point 2), there are many low distortion solid state amps that are very resolving (because distortion obscures detail) but they are also bright because while the distortion they make is very low, its highly audible to the ear because its not masked and also because the ear is keenly sensitive to their presence.
(BTW to me this is neither accurate or natural in sound despite the THD being quite low).
If there is too much bass or if the treble is lacking the system will sound muffled and less resolution.
So its important that both the frequency response be as flat as possible and the distortion benign. The tricky bit is that the ear converts distortion to tonality (hence the 'warmth' of tube amps and the brightness of solid state) and so tends to give tonality imparted in that manner more weight. For this reason you can have a significant FR error that might not get noticed if the distortion is benign.
Thank you, Ralph, for your reply.
Here when you say "less resolution" do you mean actually less resolution in the video analogy sense of fewer pixels per inch? Or do you mean only a subjective sense or impression of less apparent detail and less resolution because the treble is being de-emphasized or swamped by bass?
More interesting to me @Ron Resnick is that there is also some correlation with lower harmonic content sounding like it is lower volume, but the SPL measuring high.
And higher harmonic content sounding louder “and richer”.
Maybe some of ^that^ biology feeds into why some people like SET amps, and higher distortion speakers for lower level listening.
It also can make low distortion amps and speakers somewhat of a danger is one likes to turn them up until they sound loud.
Maybe the distortion “appears” to be more “resolving”, but those devices are putting in things that are not true to the recording… However we all know that high fidelity and what people like are not exactly the same thing.
^That^ said, I appreciate low distortion amps and speakers and systems that “sound’ somewhat quiet, when they are not quiet.
But I also pull out the iPhone/iPad NIOSH app if I think that the SPL maybe getting into the danger range.
I have also heard Ralph mention mention “loudness” and distortion being tied at the hip..
Similar threads
- Sticky
