That's an interest point.
The bodzio pulse paper linked on the first page shows how linear-phase crossovers sum to cancel each others pre and post ringing, when used fully equivalent on both sides of crossover.
But that's for electrical, and needs to hold for acoustic too...which as you say, may not hold for off-axis summation.
I get that steeper linear-phase crossovers that don't sum acoustically, could potentially exacerbate pre-ringing.
I've been using steep crossovers to minimize off-axis comb filtering, by narrowing the critical summation region. It works!
It's been giving measurably improved off-axis consistency (for live sound as well as uniform indoor power response)
What I wonder is whether a maybe pre-ringing summation issue, is worse than an easy to discern comb filtering problem.
I really doubt it, but it bears investigation if i can figure out a way to do it
Sure. I get all that
How would you go about pin-pointing the audibility of the off-axis pre-response you discussed?
That is a big undertaking... The number of room and driver combinations is significant. My advice would be to use drive units that are well-behaved out-of-band and off-axis: The rules are essentially the same as with any crossover.
FYI I have gone back to 4th and 8th order minimum phase crossovers, but still have a linear phase roll-off at low frequency. Implementing Thiele notch filters is an option worth trying, as is Hawksford's 'blending technique' (for which I have forgotten the name and the reference, sorry).
FYI I have gone back to 4th and 8th order minimum phase crossovers, but still have a linear phase roll-off at low frequency. Implementing Thiele notch filters is an option worth trying, as is Hawksford's 'blending technique' (for which I have forgotten the name and the reference, sorry).
I hear you about the undertaking....I only try to measure/listen off-axis outdoors, to help simplify.
Yes, I can see using min-phase with higher frequencies, as group delay gets pretty minimal.
It's ironic isn't it, that the real need to minimize group delay (down low) is where nearly everyone says it's inaudible (which i really disagree with)
I'm using lin-phase LR 8th order for the 3 xovers in a 4-way.
Along with ith a 3rd order BW HP at the bottom, and a 2nd order LP at the top, both lin-phase.
Been playing with moving FIR impulse centering away from middle towards the front, ....seems to help down low a little.
Yes, I can see using min-phase with higher frequencies, as group delay gets pretty minimal.
It's ironic isn't it, that the real need to minimize group delay (down low) is where nearly everyone says it's inaudible (which i really disagree with
)I'm using lin-phase LR 8th order for the 3 xovers in a 4-way.
Along with ith a 3rd order BW HP at the bottom, and a 2nd order LP at the top, both lin-phase.
Been playing with moving FIR impulse centering away from middle towards the front, ....seems to help down low a little.
If you can tell the difference between different MP3 conversion rates, you would be able to hear ringing in a filter. It seems to be a very similar phenomenon.
Aliasing and pre-responses are different phenomena. They do not sound alike.
Funny! And Hi Jack,
But I don't get how you can equate a decreased mp3 sample rate, with FIR pre-response ??? Seem like entirely different animals, no ?
Please state which filters and/or implementation has led you to come to this conclusion, so we can avoid it!
mark100 have seen some of your standing on axis nearfield acoustic slopes and honestly they are precision in comparison to most stuff and will imagine that feature bring you satisfaction and progress so that linear phase XO point filters also work as optimal and distort as less as possible. General speaking about subjective sensed conclusions is probably a hard subject to discuss and agree about in public domain where we all have some passion about our gear and solutions to get that musical material perform as close to art as possible into our rooms. To get back to what precision slopes mean can tell i nerd about these myself and probably because we can and also some DSP engines nowadays is pretty powerfull, zbig001 linked on previous page to a test about digital aliasing sound test which i ran but also took same sites DR test and surprised got resolution -60dB down in a relative noicy room so probably not a bad idea nerd about those slopes, that said its true use of linear phase filter will cause some more or less global system delay and respect its not all users or application that can live with that feature.
if preringing is audible, i would think using more taps would reduce the audibility by making it gentler. some dsps have low limits on taps and cpu and its audible imho. some allow lg. number and it sounds better to a point of diminishing returns.
ive been using lin phase xo for about eight months now. i know its better than passive but i cant be sure it has much to do with linear phase specifically. perhaps i need to spend time with min phase xos.
ive been using lin phase xo for about eight months now. i know its better than passive but i cant be sure it has much to do with linear phase specifically. perhaps i need to spend time with min phase xos.
I've not knowingly heard pre ringing. I've not knowingly listened closely to a digital system where pre ringing may exist, although ive probably listened to a source with pre ringing and not noticed.
From what I understand, a filter type with a defined pass band ripple, such as Chebychev, with exhibit more of it than a Bessel for example.
Pre ringing isn't a theoretical effect (which may or may not be audible), which would discourage me from using digital filters.
The presence (or lack of) certain features in particular DSP implementation are my primary reasons for avoiding affordable DSP.
Such as 3v3 supply rails, output voltage levels, DAC quality, these would limit my options right away. There doesn't seem to be one implementation which 'has it all and at a price I'm happy paying.
From what I understand, a filter type with a defined pass band ripple, such as Chebychev, with exhibit more of it than a Bessel for example.
Pre ringing isn't a theoretical effect (which may or may not be audible), which would discourage me from using digital filters.
The presence (or lack of) certain features in particular DSP implementation are my primary reasons for avoiding affordable DSP.
Such as 3v3 supply rails, output voltage levels, DAC quality, these would limit my options right away. There doesn't seem to be one implementation which 'has it all and at a price I'm happy paying.
This thread appears to be going off course a little from its origin concerning the audibility of pre-echo. Strictly speaking, I prefer use of the term pre-response to describe filters and reserve pre-echo for the discrete effects that do actually occur in nature: I think here of "print through" on magnetic tapes, for example.
As my first post in this thread tried to make clear, the audibility of pre-responses depends upon the application. Crossovers are applied in (ideally) complimentary guise, such that any audible effects result from errors in the sum of different acoustic paths. Anti-aliasing filters are not used in compliments and so leave their effects convolved with the signal. Also we need consider that crossovers are concentrated in the audible bandwidth, where as anti-aliasing filters (ideally) are not.
One way to answer the question of audibility is to construct a minimum phase filter response in some DSP platform. Then simply construct alongside its maximum phase counterpart (by reversing the filter). The user can then switch between the pre-response and post-response filters, and hear for themselves that: (a) that they sound different (or not); (b) the effects are greater for high filter cut-off rates (and therefore longer filters).
It would not then be difficult to also implement the linear phase version of the filter for a third comparison. Fathoming out how this conversion can be done easily should prove useful to anyone so bothered (although you will need to start with a longer minimum phase filter padded with zeros to make this possible).
As a final word of caution to the ardent experimenter, be wary of the room in which you are listening, if indeed you are.
But it would also be a good idea to not talk of standard filter alignments like Bessel and Chebychev in this discussion except maybe for magnitude response targets in the experiment design. Things get a little difficult when trying to introduce Q into discussions of linear phase filters
As my first post in this thread tried to make clear, the audibility of pre-responses depends upon the application. Crossovers are applied in (ideally) complimentary guise, such that any audible effects result from errors in the sum of different acoustic paths. Anti-aliasing filters are not used in compliments and so leave their effects convolved with the signal. Also we need consider that crossovers are concentrated in the audible bandwidth, where as anti-aliasing filters (ideally) are not.
One way to answer the question of audibility is to construct a minimum phase filter response in some DSP platform. Then simply construct alongside its maximum phase counterpart (by reversing the filter). The user can then switch between the pre-response and post-response filters, and hear for themselves that: (a) that they sound different (or not); (b) the effects are greater for high filter cut-off rates (and therefore longer filters).
It would not then be difficult to also implement the linear phase version of the filter for a third comparison. Fathoming out how this conversion can be done easily should prove useful to anyone so bothered (although you will need to start with a longer minimum phase filter padded with zeros to make this possible).
As a final word of caution to the ardent experimenter, be wary of the room in which you are listening, if indeed you are.
But it would also be a good idea to not talk of standard filter alignments like Bessel and Chebychev in this discussion except maybe for magnitude response targets in the experiment design. Things get a little difficult when trying to introduce Q into discussions of linear phase filters
Possible, recordings can surprise with ingenuity...
As to the occurrence of an audible pre-echo in music:
It's been two years since I've used linear-phase filtration in my open baffles, but in music I have not yet been able to observe such a phenomenon.
On other hand, my speakers are standing in a room that is too small for them, at high volume levels the damping elements are no longer enough and some sound nuances can be masked.
For those interested generally in the subject of FIR filtration, I recommend the rationalaudiophile blog.
The blog is rich both in deeply detailed analysis of the subject from technical side, as well as very insightful descriptions of the listening experience.
At least as far back as Peter Gabriel in the mid 80s and used extensively in the industry thereafter... Maybe someone knows the history better?
You may find interesting some of the work by Rob Watts, consultant engineer to Chord Electronics, developing the Hugo M Scaler. Alias frequencies, pre and post ringing, and transient distortion in the ADC and DAC processing, issues seriously attempting to be addressed. One of Watts' concerns is that the ear/brain is able to resolve up to 4us, yet CD Redbook resolution is 22us (44 kHz). He plains to address analogue to digital conversion issues in the future Chord Davina ADC.
Rob Watts' video presentation of the Hugo M Scaler at the CanJam in 2018.
YouTube
Post #23 in the thread for Rob Watt's Powerpoint presentation of the Hugo M Scaler
Hugo M Scaler by Chord Electronics - The Official Thread | Page 2 | Headphone Reviews and Discussion - Head-Fi.org
Post #2072 in the thread, attaching some of Rob Watt's comments to the CanJam H M Scaler Powerpoint presentation.
Hugo M Scaler by Chord Electronics - The Official Thread | Page 139 | Headphone Reviews and Discussion - Head-Fi.org
Post #2168 in the thread, Rob Watts summarizing some design goals of the H M Scaler
Hugo M Scaler by Chord Electronics - The Official Thread | Page 145 | Headphone Reviews and Discussion - Head-Fi.org
Rob Watts' video presentation of the Hugo M Scaler at the CanJam in 2018.
YouTube
Post #23 in the thread for Rob Watt's Powerpoint presentation of the Hugo M Scaler
Hugo M Scaler by Chord Electronics - The Official Thread | Page 2 | Headphone Reviews and Discussion - Head-Fi.org
Post #2072 in the thread, attaching some of Rob Watt's comments to the CanJam H M Scaler Powerpoint presentation.
Hugo M Scaler by Chord Electronics - The Official Thread | Page 139 | Headphone Reviews and Discussion - Head-Fi.org
Post #2168 in the thread, Rob Watts summarizing some design goals of the H M Scaler
Hugo M Scaler by Chord Electronics - The Official Thread | Page 145 | Headphone Reviews and Discussion - Head-Fi.org
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.