But as above, sometimes it is useful
The style in which you did it yes. It is when the entire post is quoted as a block and then responded to (and that exact same block of text is directly above) that it is just causing clutter.
Tony.
Ooops, are you describing (A) the audio world you wish we lived in or (B) the real audio world where the phases spin through the air (and spin down your cochlear organ) and can't be manipulated (by any feasible means today) into arriving at your ears in phase (in-canal headphones excepted)?
Link appreciated.
B.
It is all in the maths - or rather in the measurements. Just for the most part, simple frequency responses often blur the bits that are important. Using a magnitude response, for example, which is a second order mapping, removes the phase information altogether. Hence we might need look at a third order measure such as the bispectrum. Far more fundamental than identifying the effects of phase linearisation, here you might also find the means by which you identify a given instrument or recognise a person talking to you on the telephone, for example.
You can manipulate the phase anyway you like by pre-processing. I am not sure I understand the question...
There are three threads at least I can think of that I have contributed to on the subject:
What is the point of a centre speaker?
Fixing the Stereo Phantom Center
Mid-side stereo techniques
Ok, try looking at the parts separately, starting with the case of the delay. If you collected that greatly rolled around phase into a measurement program and subtracted the delay, the phase would look good again. Apparently our ears can work this out. They say it is something we are instinctively attuned to. I've heard it from PhDs. My own experience confirms it, and I go with it. So I also go with the group delay threshhold data regarding intra spectrum shifts. I used to run a waveguide 30cm behind a woofer, crossed at 1k2 and compensated using an asymmetrical passive crossover. The anomaly was on the border of being inaudible.
The reverberant field, sound which has gone past my head and has returned. I want it to be as varied in time, distance and angle as it can be so it is a smooth and indistinct background presence. When things begin lining up you get loud spots and distraction. The field should roll by smoothly.
Then look at the direct field. It's integrity is responsible for the distributions in the first place. It can to some degree be looked at as a geometric issue. If you get a particular reflection that has a unique character it can stand out and fatigue you. The more it sounds like it belongs to the direct sound, the less your brain has to try to tune it out and that is the basis of listening fatigue. A good configuration shouldn't tire even after hours of listening. It doesn't stop there, because it will also lead to a failure to properly set up the reverberant field.
This is far too presecriptive. The effect of a reflection is very dependent on the time it arrives after the direct arrival and also on its direction. And the definition of a reverberant field precludes discrimination of its source. But we are veering off-topic...
?? could you perhaps clarify this statement? while i can agree that reverb can affect localization i can't see it precluding localization of a source, even in conditions of multiple echoes from different directions we can extrapolate the source (provided it's not in an acoustic shadow)
Once you have established a reverberant field, it is not possible to discern the individual reflections that led to it. If you can discern individual reflections then it is not a reverberant field. But we are veering off-topic...
This thread has fallen so far afield of the original question I don't feel bad about this minor hijack.
Some of the issues that you touch on above are dear to me, specifically the reverberent field and then what you might call the power response of the system, which will have a direct effect on the tonal character of the reverberent field. Case in point a loudspeaker using both horn loaded and direct radiator drivers, like a PA speaker, but even a speaker with a "waveguide" may sometimes fall into this category. I find many speakers with true horn loaded upper frequencies to sound terrible except at one particular spot in the room where the on-axis SPL is as far above the reverberant (e.g. room) SPL as possible. Often these speakers sound best toed in, in a dead room with lots of sound treatment, and in a rather confined spot. Move off axis and the very uneven power response makes itself known very quickly. The sound is dead, muddy. It's not for me.
This is why I prefer relatively live rooms with mostly bare walls that provide strong reflections across the audio spectrum, combined with a type of speaker that has as flat of a power response as possible, meaning that the off axis frequency response looks very much like the on axis response. To mean this has come to mean a dipole loudspeaker. You might think of omni types, as doing this also but I don't have much experience with them.
If the response looks very much the same on all axes (at least in the horizontal plane) then you do not need to worry about "treating" the room. It will simply return some delayed copies of the speaker output more or less. The brain is well adapted to fold these into the direct sound, and there is some evidence (both theoretical and experiential) that this "enhances" the cues within the recorded material (thinking stereo only here) and causes the listener to hear more depth and detail.
It matters not whether the reflections are "strong" or "diffuse" or "weak". What matters is their tonal balance WRT the on axis sound. When the on and off axis acoustic radiation have widely varied frequency responses, you need to put a "band aid" on the reflected sounds using diffusers or adsorption, and people go to great lengths down this path when they could just get it right from the beginning with a better type of loudspeaker that will interact in a better way with the room.
Last edited:
I disagree! The thread concerned characterising a "transient response". It has explored the linear system theory relevant to the impulse response and the discerning of audible phenomena that might be considered to contribute to the perception of a "transient". It has been notably devoid of any answers, however.
I'll give a 2 cent stab at real world too...
First thing...what is phase anyway?
Simply.... the relative timing between waveforms (frequencies) combining to make sound.
Relative timing which is expressed as degrees rotation by frequency.
Ok, we all know all that...but i think we often forget whenever we use the word phase, we should remember it means relative phase, relative timing.
And that relative phase, that leaves the speaker stays the same wherever it arrives, at whatever the distance.
So the idea that phase gets all jumbled just because it travels some listening distance doesn't hold water.
(It does get jumbled from reflections and reverb, but that is subsequent combining has nothing to do with whether phase matters or not for the initial radiation.)
For me, that isn't theoretical textbook stuff...it's just common sense.
I mean all sound is, ...are vibrations (frequencies) commingled and the relative timing between those vibrations.
How the heck can relative timing (phase) not matter????
The question for me and i guess everyone, is how much does it matter.
I dunno, and who can say definitively...
But i do know as I've managed to get phase flatter and flatter, i keep hearing new things in very familiar tracks...understanding vocals I couldn't catch before, hearing breath, vocal inflections, picking out instruments more easily, etc,
And transients...simply wow!
(by getting phase flatter and flatter I mean both on and off axis....as getting it perfectly flat on-axis alone, is trivial with FIR, and can be surprisingly hit or miss if over corrected)
Ben, i know you are a fan of electrostats...I am too. Acoustat-X in my bedroom, and CLS stored in a closet.
Are yours full range, without any crossovers inducing phase rotation? Or at least full range other than a sub ?
My Acoustat use the high voltage tube amp and have surprising strong bass, no sub needed.
No crossover at all, and in a way, both a vertical and horizontal single driver line array.. A lot of the phase and timing issues found with conventional speakers, just don't exist with stats.
How far do you sit from yours? What do you think the ratio of direct to reflected sound is?
I think that ratio is higher with electrostats than any other speaker type I've heard.....the way HF radiates perpendicular off the panels and nearly ceases above their height is amazing.
My guess you are probably hearing better phase than you think, despite what measurements might show.
I know stats measurements are difficult, and often look messy. Someday I'm going to haul one outside and measure.
Anyway, there's no question in my mind that i'm getting better and better sound as my tuning skills evolve.
I stick with flat phase (and mag other than house curves) because it just keeps working, sounding better, no matter what new type speaker build i try.
Can i say exactly how much of the sonic improvement is individually due to improved mag, phase, or pattern control......no.
But i don't really care...cause it works. l
Linear phase is easier to work with than not, and each piece of that tuning set makes common sense to me.
That's my real world anyway
First thing...what is phase anyway?
Simply.... the relative timing between waveforms (frequencies) combining to make sound.
Relative timing which is expressed as degrees rotation by frequency.
Ok, we all know all that...but i think we often forget whenever we use the word phase, we should remember it means relative phase, relative timing.
And that relative phase, that leaves the speaker stays the same wherever it arrives, at whatever the distance.
So the idea that phase gets all jumbled just because it travels some listening distance doesn't hold water.
(It does get jumbled from reflections and reverb, but that is subsequent combining has nothing to do with whether phase matters or not for the initial radiation.)
For me, that isn't theoretical textbook stuff...it's just common sense.
I mean all sound is, ...are vibrations (frequencies) commingled and the relative timing between those vibrations.
How the heck can relative timing (phase) not matter????
The question for me and i guess everyone, is how much does it matter.
I dunno, and who can say definitively...
But i do know as I've managed to get phase flatter and flatter, i keep hearing new things in very familiar tracks...understanding vocals I couldn't catch before, hearing breath, vocal inflections, picking out instruments more easily, etc,
And transients...simply wow!
(by getting phase flatter and flatter I mean both on and off axis....as getting it perfectly flat on-axis alone, is trivial with FIR, and can be surprisingly hit or miss if over corrected)
Ben, i know you are a fan of electrostats...I am too. Acoustat-X in my bedroom, and CLS stored in a closet.
Are yours full range, without any crossovers inducing phase rotation? Or at least full range other than a sub ?
My Acoustat use the high voltage tube amp and have surprising strong bass, no sub needed.
No crossover at all, and in a way, both a vertical and horizontal single driver line array.. A lot of the phase and timing issues found with conventional speakers, just don't exist with stats.
How far do you sit from yours? What do you think the ratio of direct to reflected sound is?
I think that ratio is higher with electrostats than any other speaker type I've heard.....the way HF radiates perpendicular off the panels and nearly ceases above their height is amazing.
My guess you are probably hearing better phase than you think, despite what measurements might show.
I know stats measurements are difficult, and often look messy. Someday I'm going to haul one outside and measure.
Anyway, there's no question in my mind that i'm getting better and better sound as my tuning skills evolve.
I stick with flat phase (and mag other than house curves) because it just keeps working, sounding better, no matter what new type speaker build i try.
Can i say exactly how much of the sonic improvement is individually due to improved mag, phase, or pattern control......no.
But i don't really care...cause it works. l
Linear phase is easier to work with than not, and each piece of that tuning set makes common sense to me.
That's my real world anyway
Yes, but I'm not trying to write a book at the moment
. I had been following Bens journey, his comments and those of Earl outside this thread and thought I was going along with the spirit of the correct topic. I guess I'll wait and see what Ben thinks.I did not mean to dissuade any comments and the effects of early room reflections are undoubtedly related to what we perceive - and necessary therefore to include if we wish to turn our measurements into something that shows what we perceive. Since early reflections can "fuse" with the direct source, they need to be accounted for in our transient perception. My comments were aimed to move considerations from the reverberant end of the timescale - where in essence we appear to find initial time resolution traded for frequency resolution. Hence steady state models are not the end of the story.
Not at all I agree that there is a bigger picture, and also that part of it is understanding what is audible and why. Much of my speaker building these days comes down to compromises and threshholds. Audibility criteria makes things possible.
In this respect the goals of narrow or wide constant directivity speakers are largely the same. In each case the speaker is carefully designed for this task. A dipole is not just an open baffle, it is a very special and complex one.
No. This thread started with a discussion of linear system analysis and morphed into questions concerning perception and, in particular, the effects of phase changes. What we then seek from a measure is the audibility of phase changes.
Last edited:
As I have failed to prevent this thread drifting off-topic, I will also add that when considering the effects of directivity, we also need consider the effect of stereo (for example). Predicting the perception of one or more sound sources via two-channel encoding and a two or more loudspeaker reproduction system is not trivial.
I believe mark100's reasoning is faulty. And there is a motif in recent posts that can be called "reconstruction" or "reverse engineering" of phases by the brain or maybe "wishful thinking". For example, the discussion of echoes.
The phase of separate frequencies of the glockenspiel strike spin through the air to the mic and then later to your ear and then along the length of your cochlear organ. The rotation may be orderly in a math sense, granting mark100 that. But what does it "look" like to the ear? Is the brain supposed to guess what the original sound is like? That's a logical fallacy of "reconstruction" that occurs whenever you assume the original source signal is faithful in phase to the glockenspiel in the first place and all you need to do is get your XO perfect for phase.
No doubt mark100 has cultivated hearing and taken great care in developing his audio environment. But the process of making his phases do his bidding certainly also entrained all kinds of other audible tweaks along the way unrelated to phase per se.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.