Uniform phase shifts in all channels are rather inaudible. Interaural phase shifts are audible and fundamental to spatial hearing.
One can only EQ minimum phase signals, not non-minimum phase signals. Meaning one EQ the direct signal and sometimes the very lowest frequencies (normally below 100 Hz). Or where the measured response is not comprised of superposed (combined) direct and indirect signals from the room and the speaker
EQing non-minimum phase signals is a poor choice and never works with a good effect because there's no proper correlation between phase and amplitude. Better to get well designed speakers and do acoustic treatment.
EQing non-minimum phase signals is a poor choice and never works with a good effect because there's no proper correlation between phase and amplitude. Better to get well designed speakers and do acoustic treatment.
Dirac explains when and why a mixed phase filter can work:
http://www.dirac.se/media/12044/on_room_correction.pdf
If the discrepancies between 3Khz and 7Khz are due to the mic being repositioned between measurements then isn't that a good example of why EQ'ing using high frequency measurements that include room reflections is wrong ?
Even if those reflections affect what we hear to some degree, (and I'm not convinced that they do, at least for tonal balance) the fact that the measurement can vary so much with small movements of the microphone makes it impossible to use as something to EQ against.
Mic position A and position B a number of inches apart give two very different responses and result in two very different sets of corrective PEQ's - which one is correct ? Neither! 😉
On the other hand a truly reflection free windowed measurement of the speakers high frequency response (here high frequency means above 2Khz) will not vary significantly if the microphone is moved around a few inches on the horizontal plane. (unless the speakers have bad diffraction problems, but that's a separate issue to be addressed)
I really do firmly believe that we should not be applying any high Q PEQ's at high frequencies to correct the room response, only the speaker's own reflection free response.
A little bit of broad, subtle shelving or gradual tapering perhaps, but not narrow band high Q corrections. Every time I've tried that it sounds bad.
With few exceptions all multi-way speakers are decidedly NOT minimum phase. So you can't just correct any error in a speakers response with pre-crossover minimum phase PEQ.
But you can correct errors that are well away from crossover frequencies completely. Say your mid to tweeter crossover is 2Khz, and you have a correctable tweeter resonance at 8Khz - you can correct that with a PEQ applied before the crossover because only the one driver has significant output at that frequency. If the resonance was within an octave of 2Khz you could correct the amplitude response but not the phase. Still an improvement in most cases though. (although you'd be better to fix that problem in the crossover design)
Last edited:
I've shown ungated measurements hence the large differences. We don't perceive them. But we don't perceive the free field response either. The question is which reflections to include and which ones not.
I agree; coherent (phase corrected) loudspeakers.
_________________
* Remember this too: Each time you EQ the sound, you EQ for one specific set of ears, so it cannot be definitive to all ears.
..And what you EQ first is the less than perfect original music recording, and that, is treacherous territory @ best.
^
We need to distinguish between equalizing sounds in production which is solely preference based and equalization used in reproduction which tries to remove unwanted distortion caused by the room and speaker. This thread is about the latter.
We need to distinguish between equalizing sounds in production which is solely preference based and equalization used in reproduction which tries to remove unwanted distortion caused by the room and speaker. This thread is about the latter.
Simon
I tend to agree with you here. But I think there can be exceptions. If one does not control the very early reflections then EQing to the first few ms in-situ can be a significant improvement. I have done this in an automobile setting before. But ideally the speakers should not need EQ of the direct field and they and the room should be setup such that the reflections are later than say 10 ms. When this is done then EQ above the modal region is never going to be a good thing. Only in situations that have serious problems will higher frequency EQ be an improvement.
Markus - I started reading the Dirac paper, and I mostly agreed with what I read, but I didn't read it all because its all very well known.
Very true, and the equalized room's acoustics, with our preferred target curve, will reflect the original music recording; exposing it all with the goods and bad production mixes.
With time we'll learn how to listen to only the right quality recordings of the music genres we love. ...Recorded by the real expert music recording/mixing engineers in the right studios and venues from the good music record labels.
And the distortion we try to remove from poor room's acoustics and bad speaker's dispersion by equalizing with a parametric EQ has to come from a playing music source. And that source will determine the overall amount of EQ at the end.
What's the point of having a good equalized room if the source is poor to start with?
The music playing is the ultimate test, not the measurements.
Last edited:
Some music recordings will be accommodated better than others in a particular equalized room.
The room's acoustics reproduced by the speaker's sound dispersion in that room and using a parametric equalizer or several of them to correct the dips and peaks, won't be the all-in-one solution for all music recordings. ...That's what I'm saying.
But yes, that is the best we can do for now. ...Till we have some EQs that can automatically adjust for each recording in correlation with the room's acoustics.
The music playing has its own signature, one that creates with the room and the speakers a totality of sound dispersion.
And even the overall volume level has major influences, and for each album, and sometimes even for each tune (song, track).
_____________
Post some cool graphs with some nice looking room's equalized curves created from parametric equalizers, and try to replicate them in another room, and describe them in real life music playing; is part of the goal. The other part is to make them pleasing for one person, and then for everyone.
And that is a challenging proposition, with a parametric equalizer, or two.
The room's acoustics reproduced by the speaker's sound dispersion in that room and using a parametric equalizer or several of them to correct the dips and peaks, won't be the all-in-one solution for all music recordings. ...That's what I'm saying.
But yes, that is the best we can do for now. ...Till we have some EQs that can automatically adjust for each recording in correlation with the room's acoustics.
The music playing has its own signature, one that creates with the room and the speakers a totality of sound dispersion.
And even the overall volume level has major influences, and for each album, and sometimes even for each tune (song, track).
_____________
Post some cool graphs with some nice looking room's equalized curves created from parametric equalizers, and try to replicate them in another room, and describe them in real life music playing; is part of the goal. The other part is to make them pleasing for one person, and then for everyone.
And that is a challenging proposition, with a parametric equalizer, or two.
Last edited:
It's the essence of life itself man, don't you dig it!
😛* Just click my sig, you'll see why!
Last edited:
Earl, read through this forum. How many people do you think understand the difference between terms like minimum phase and mixed phase? I've linked the paper to bring those people up to speed.
Last edited:
How many people have listened for an extended period of time to compare min, mixed, and linear phase FIR filters for digital XO, time alignment, driver linearization, and room correction?
I have come to the same conclusion as Arthur has with respect to FIR correction filter type. Quoted from: The Building of a Pure Digital System
"I would describe the sound of my system “real”. It has a wide soundstage, ultra high resolution with excellent micro and macro dynamics. The timber, the texture and layering of each musical instrument are the best that I have heard. Violin sounds like real violin and piano sounds like piano. In orchestral musical piece, I can hear each individual equipment clearly in their own location in space. Much of this is brought about by the use of linear phase filter and phase alignment. With minimal phase filter, (typical passive crossover) you feel that all instruments are blended together and their relative position is not as defined. You can only hear roughly different sections, the individual instrument within each section cannot be defined. But if you have the experience in attending life classical concert, you will notice that this is not the case. You close your eyes and you should be able to tell exactly where is the instrument. Think about it, the conductor definitely knows which player is off the pace and the recording microphone is frequently placed just above him!"
I think most folks would be nothing short of astonished on how sophisticated DSP software like Acourate and the power of the PC can improve ones existing speakers and listening environment with:
- Linear phase FIR correction filters at 64 bit resolution with 65,536 taps utilizing less than 1% CPU on a midrange PC.
- FIR filter contains independent amplitude and phase room correction for any number of channels, and optionally, digital XO, time alignment, and individual driver linearization.
- a psychoacoustic filter that matches closely to what our ears perceive in small acoustic environments - I have yet to find any other software with this feature.
- frequency dependent windowing (FDW) analysis so one is not measuring just a single point in time, especially in the bass frequencies where the window is progressively getting longer with decreasing frequency. On the top end, constant directivity waveguides ensure good coverage of the listening area with the same spectral balance, standing or sitting. Both achieved with one mic measurement position.
- standardized Interaural Coherence Coefficient (IACC) measure which is a measure of channel and room reflection equality for the first 80 milliseconds of sound travel. Think of sweeping slowly from 20 Hz to 20 kHz with the tone never wavering in perceived amplitude or stereo imaging, it is always in the center, and for a finite period of time. Think of what that means from an imaging perspective, even if your stereo is not symmetrically aligned. Try a similar test Online LEDR Sound Test | Listening Environment Diagnostic Recording Test to hear how your stereo system images.
- customized target frequency response to a standard or your own preference. There is a standard frequency response target that yields a *perceptually* flat frequency response at the listening position, independent of speakers and room and works for virtually every music category.
Listening to this digital correction, even on my less than speakers and room, as far as I have been able to research and listen to, is the state of the art in room correction, digital XO, time alignment, and driver linearization.
Having played with PEQ (analog and digital) for room correction for decades, I would not compare it as being in the same category as today's dedicated digital loudspeaker and room correction software.
For me, the more I can match the digital waveform (i.e. music stored on disk) to the acoustic waveform arriving at my ears at the listening position, the more accurate the reproduction of the spectral balance and stereo imaging is. Basically decoding what was encoded by the mixing, mastering engineers, producers, artists in the control room when it was mixed/mastered. Source: recording/mixing engineer for ten years.
Hearing a properly measured and corrected system, using purpose built DSP software, I would think most people would be surprised at the level of improvement without any other changes to ones existing system. Try it for yourself. If anyone finds anything better, please let me know.
Hope that helps. Best regards, Mitch
I have come to the same conclusion as Arthur has with respect to FIR correction filter type. Quoted from: The Building of a Pure Digital System
"I would describe the sound of my system “real”. It has a wide soundstage, ultra high resolution with excellent micro and macro dynamics. The timber, the texture and layering of each musical instrument are the best that I have heard. Violin sounds like real violin and piano sounds like piano. In orchestral musical piece, I can hear each individual equipment clearly in their own location in space. Much of this is brought about by the use of linear phase filter and phase alignment. With minimal phase filter, (typical passive crossover) you feel that all instruments are blended together and their relative position is not as defined. You can only hear roughly different sections, the individual instrument within each section cannot be defined. But if you have the experience in attending life classical concert, you will notice that this is not the case. You close your eyes and you should be able to tell exactly where is the instrument. Think about it, the conductor definitely knows which player is off the pace and the recording microphone is frequently placed just above him!"
I think most folks would be nothing short of astonished on how sophisticated DSP software like Acourate and the power of the PC can improve ones existing speakers and listening environment with:
- Linear phase FIR correction filters at 64 bit resolution with 65,536 taps utilizing less than 1% CPU on a midrange PC.
- FIR filter contains independent amplitude and phase room correction for any number of channels, and optionally, digital XO, time alignment, and individual driver linearization.
- a psychoacoustic filter that matches closely to what our ears perceive in small acoustic environments - I have yet to find any other software with this feature.
- frequency dependent windowing (FDW) analysis so one is not measuring just a single point in time, especially in the bass frequencies where the window is progressively getting longer with decreasing frequency. On the top end, constant directivity waveguides ensure good coverage of the listening area with the same spectral balance, standing or sitting. Both achieved with one mic measurement position.
- standardized Interaural Coherence Coefficient (IACC) measure which is a measure of channel and room reflection equality for the first 80 milliseconds of sound travel. Think of sweeping slowly from 20 Hz to 20 kHz with the tone never wavering in perceived amplitude or stereo imaging, it is always in the center, and for a finite period of time. Think of what that means from an imaging perspective, even if your stereo is not symmetrically aligned. Try a similar test Online LEDR Sound Test | Listening Environment Diagnostic Recording Test to hear how your stereo system images.
- customized target frequency response to a standard or your own preference. There is a standard frequency response target that yields a *perceptually* flat frequency response at the listening position, independent of speakers and room and works for virtually every music category.
Listening to this digital correction, even on my less than speakers and room, as far as I have been able to research and listen to, is the state of the art in room correction, digital XO, time alignment, and driver linearization.
Having played with PEQ (analog and digital) for room correction for decades, I would not compare it as being in the same category as today's dedicated digital loudspeaker and room correction software.
For me, the more I can match the digital waveform (i.e. music stored on disk) to the acoustic waveform arriving at my ears at the listening position, the more accurate the reproduction of the spectral balance and stereo imaging is. Basically decoding what was encoded by the mixing, mastering engineers, producers, artists in the control room when it was mixed/mastered. Source: recording/mixing engineer for ten years.
Hearing a properly measured and corrected system, using purpose built DSP software, I would think most people would be surprised at the level of improvement without any other changes to ones existing system. Try it for yourself. If anyone finds anything better, please let me know.
Hope that helps. Best regards, Mitch
- Status
- Not open for further replies.