Hmmm, it depends. With clean, low distortion loudspeakers, it is VERY difficult (if not impossible) to detect absolute polarity on musical program. It can be detected on certain test signals. With loudspeakers having higher distortion (especially high second order), detection of polarity is quite a bit easier.
imho there should not be any significant phase shift within audible band because otherwise signal reproduction would be affected badly. Coherency of a signal is a very audible thing.
BBE claims to adjust signal in advance to correct most imperfections that would affect signal phase later on in amplifier-speaker in a way so they cancel each other to produce perfect result finally.
Form my experience BBE does not work as originally intended. Оn a good setup it spoils sound so it is better be switched off. On an entry level setup I cannot hear any benefits from BBE anyway due to low audio tract resolution.
All above is very subjective of course.
BBE claims to adjust signal in advance to correct most imperfections that would affect signal phase later on in amplifier-speaker in a way so they cancel each other to produce perfect result finally.
Form my experience BBE does not work as originally intended. Оn a good setup it spoils sound so it is better be switched off. On an entry level setup I cannot hear any benefits from BBE anyway due to low audio tract resolution.
All above is very subjective of course.
It's Black Or White.....
SY, you got any citations on that...please enlighten us - I think you have that the wrong way round.
In my oh so limited listening experience, clean sounding single cone loudspeakers are the most revealing of absolute polarity, and typical multiway loudspeakers are typically the least capable of differentiating absolute polarity.
Speakers exhibiting high second order distortion at high/overload levels will give rise to a polarity preference, but this is not necessarily correct indication of repro-chain correct absolute polarity.
Correct absolute polarity conveys sounds of the real world and any aberrations (including frequency dependent delays) will confuse this portrayal of the illusion that is called 'stereo'.
Correct polarity sounds and inverted polarity sounds are two different animals.
Eric.
SY, you got any citations on that...please enlighten us - I think you have that the wrong way round.
In my oh so limited listening experience, clean sounding single cone loudspeakers are the most revealing of absolute polarity, and typical multiway loudspeakers are typically the least capable of differentiating absolute polarity.
Speakers exhibiting high second order distortion at high/overload levels will give rise to a polarity preference, but this is not necessarily correct indication of repro-chain correct absolute polarity.
Correct absolute polarity conveys sounds of the real world and any aberrations (including frequency dependent delays) will confuse this portrayal of the illusion that is called 'stereo'.
Correct polarity sounds and inverted polarity sounds are two different animals.
Eric.
Last edited:
Ok, a 4" or 5" that does vocals nicely and feed it spoken word.
Get a friend to stand in open air outdoors and speak directly toward you or directly away from you...repeat the experiment indoors.
In my experience you will have four different sounds....the clue is the polarity of the first impulse and relative maxima of half waves....this denotes the polarity of energy direction.
A decent vocal range loudspeaker will portray vocal range signal polarity states easily.
Eric.
Get a friend to stand in open air outdoors and speak directly toward you or directly away from you...repeat the experiment indoors.
In my experience you will have four different sounds....the clue is the polarity of the first impulse and relative maxima of half waves....this denotes the polarity of energy direction.
A decent vocal range loudspeaker will portray vocal range signal polarity states easily.
Eric.
Last edited:
he he, maybe better like that:
- we use the interaural arrival time difference for localizing a source emitting under 1500 Hz, can be named phase shift between the two ears
- BUT, this doesn't establish that we are ipso facto sensitive to the phase shift affecting the components of a signal.
- at high frequencies, over 1500 Hz (but with some blank zones like 2000 ), then we use the interaural level difference, the source is perceived on the louder side.
A single cone speaker will suffer much more from Doppler distortion. In order to cope with HF it will need to be relatively small, so it will need higher displacement at LF so more distortion from suspension non-linearities. Did anybody say "euphonic colourations"?
It makes sense that polarity is easier to detect when second-order distortion is present - this will help create a reference which can be compared with inherent second-order components (e.g. inherent in the ear).
It makes sense that polarity is easier to detect when second-order distortion is present - this will help create a reference which can be compared with inherent second-order components (e.g. inherent in the ear).
Four different sounds, yes, but not for the reason you claim. The polarity of the initial impulse will actually be the same in all cases. What will be different is the frequency response, and reverberation.
Exactly, that was my point. Intrachannel phase information is what creates the sound stage, stereo image, depth of field, etc - all of the elements that combined give rise to an illusion of a musical performance in our living room (I don't listen in mono). Is this not what we want from our playback systems? If an FFT plot cannot show this sort of information or discern differences in this area then how useful is it in evaluating differences of this nature between different systems or modifications to existing systems?
In other words are FFT plots a castrated tool & any claims that it is showing us all there is to see is being made by eunuchs?
Last edited:
Of course they can, if that's the measurement you've set up.
I'm risking another outburst here, but I have to repeat, you're not getting what an FFT is- it is NOT a measurement, it's a way of manipulating data (usually time sequence data). You can FFT any sort of data, the question is really, for a given thing one is trying to measure, how does one get the data to FFT? If you're looking for interchannel phase differences, then you want to examine the output of a differential measurement.
I'm risking another outburst here, but I have to repeat, you're not getting what an FFT is- it is NOT a measurement, it's a way of manipulating data (usually time sequence data). You can FFT any sort of data, the question is really, for a given thing one is trying to measure, how does one get the data to FFT? If you're looking for interchannel phase differences, then you want to examine the output of a differential measurement.
FFT can show this information. Simple procedure:
1. get the phase info from FFT of left channel.
2. get the phase info from FFT of right channel.
3. subtract to get phase difference.
Alternatively, get the FFT results in complex number form (as the FFT actually produced them). Divide one channel by the other. Any deviation from unity is an inter-channel difference. What is all the fuss about? Does the OP know the difference between an FFT (i.e. the Fast Fourier Transform algorithm) and the spectrum amplitude plot sometimes called an FFT? As I have said, an FFT tells you all you need to know about steady-state response under linear conditions, and in most cases also all you need to know about transients too. You do need to understand complex numbers, of course, but that is true for most of analogue electronics. An awareness of the power and limitations of Fourier theory helps too.
1. get the phase info from FFT of left channel.
2. get the phase info from FFT of right channel.
3. subtract to get phase difference.
Alternatively, get the FFT results in complex number form (as the FFT actually produced them). Divide one channel by the other. Any deviation from unity is an inter-channel difference. What is all the fuss about? Does the OP know the difference between an FFT (i.e. the Fast Fourier Transform algorithm) and the spectrum amplitude plot sometimes called an FFT? As I have said, an FFT tells you all you need to know about steady-state response under linear conditions, and in most cases also all you need to know about transients too. You do need to understand complex numbers, of course, but that is true for most of analogue electronics. An awareness of the power and limitations of Fourier theory helps too.
if it talks like a Troll...?
no one on the engineering side is claiming "...FFT.. ..is showing us all there is to see..." - we use it as a particularly powerful tool for viewing some aspects of systems with some classes of test signals - your statement is a classic "strawman" exageration/rhetorical attack and the reason you're being called a troll
and even quite ironically entertaining as pointed out FFT is one of the more powerful relative phase measurement tools
for a look a little deeper into FFT measurement applications you could look up "fast convolution" - often used for "transient" analysis
no one on the engineering side is claiming "...FFT.. ..is showing us all there is to see..." - we use it as a particularly powerful tool for viewing some aspects of systems with some classes of test signals - your statement is a classic "strawman" exageration/rhetorical attack and the reason you're being called a troll
and even quite ironically entertaining as pointed out FFT is one of the more powerful relative phase measurement tools
for a look a little deeper into FFT measurement applications you could look up "fast convolution" - often used for "transient" analysis
Last edited:
Juts my very primitive understanding of a subject please correct me where I am wrong if somebody would have some free time an willing to do so.
FFT is a subclass of Fourier transform because mathematicians operates plus minus infinity or N+1 tha goes to infinity easily but in real life plus minus infinity is not so practical
. So FFT is not 100% accurate but easily doable in real life and gives pretty decent approximation for audio signals actually.
In general Fourier transform just an interpretation of a complicated hardly to understand from scratch function with a composition of well know simple co-sines. Altogether these co-sines will form us original function.
Now instead of feeding original complicated function we can feed into amp-speaker all these co-signes instead. If amp amplifies and speakers radiates all these co-sines equally (that we can be sure about based on frequency response plot) and phase shift is equal for an entire audio band (we can get it from phase plot) and there will be now extra co-sines generated by amp (FFT plot of 1KHz input for example) then we can be happy getting amplified co-sines that would form original complicated signal with multiplied amplitude. Altogether all that could be described by THD parameter imho.
FFT is a subclass of Fourier transform because mathematicians operates plus minus infinity or N+1 tha goes to infinity easily but in real life plus minus infinity is not so practical
. So FFT is not 100% accurate but easily doable in real life and gives pretty decent approximation for audio signals actually.In general Fourier transform just an interpretation of a complicated hardly to understand from scratch function with a composition of well know simple co-sines. Altogether these co-sines will form us original function.
Now instead of feeding original complicated function we can feed into amp-speaker all these co-signes instead. If amp amplifies and speakers radiates all these co-sines equally (that we can be sure about based on frequency response plot) and phase shift is equal for an entire audio band (we can get it from phase plot) and there will be now extra co-sines generated by amp (FFT plot of 1KHz input for example) then we can be happy getting amplified co-sines that would form original complicated signal with multiplied amplitude. Altogether all that could be described by THD parameter imho.
we don't have to "perfectly" represent a series of infinite sines to reproduce real music performances - which don't have "infinite" duration (subjective impressions of high school band concerts aside)
as we keep repeating FFT is a tool for processing data which can make some things easier to measure
below noise and a sine test signal are filtered, compared in LTspice - which gives cursor based phase, magnitude difference measurements
which display/cursor reading would you prefer to use to measure ampitude, phase differences at 1 KHz in this instance?
and again - phase isn't commonly displayed because the FFT "noise floor" also has full amplitude "noisey" phase variations that confuse the eye - we are usally interested in the difference in phase between a few frequencies which have higher amplitude - so navigating with the cursors - or writing a program to extract the information in a particular test is not often a limitation
as we keep repeating FFT is a tool for processing data which can make some things easier to measure
below noise and a sine test signal are filtered, compared in LTspice - which gives cursor based phase, magnitude difference measurements
which display/cursor reading would you prefer to use to measure ampitude, phase differences at 1 KHz in this instance?
and again - phase isn't commonly displayed because the FFT "noise floor" also has full amplitude "noisey" phase variations that confuse the eye - we are usally interested in the difference in phase between a few frequencies which have higher amplitude - so navigating with the cursors - or writing a program to extract the information in a particular test is not often a limitation
Attachments
Last edited:
So you are looking for - what? Phase differences to show why one device sounds different from another? And before we go any further - what type of signal do you want to analyze?
I'm citing this as an example of someone who used this FFT as shown & came to the conclusion quoted. It's not up to me to state what I want analysed - it's incumbent on those who make claims like " if it's in the analogue signal I will find it". I am stating that as per the discussion so far it would appear that this FFT plot simply is not enough to characterise a device & make the statement that was made. Are you saying that it is?
- Status
- Not open for further replies.