If the input and output sample rates are the same then SXQ does not convert the sample rate of course. I tend to use 48k\48k because that sample rate is always supported by adapters. And for the distortion problem, I`m pretty sure that its due to that adapter\driver. I get perfect results with a creative USB 5.1, best sound ever. Even though it may not seem to be hyped as high end, Creative drivers\integration is always tip top in my experience,and I had problems in the past, also with the supposedly high ended M Audio stuff, their drivers were really notorious. I dont think they exist anymore, not surprisingly.
Thank you Deodato for all the support!
I installed the KB fix for Windows. Closed all Asus Xonar software. Set Cable and Xonar both to 44.1. The ringing is 99% gone.
I think it has to do with Asus's software, even though the DTS was disabled, after closing all processes from Asus it is all OK.
I installed the KB fix for Windows. Closed all Asus Xonar software. Set Cable and Xonar both to 44.1. The ringing is 99% gone.
I think it has to do with Asus's software, even though the DTS was disabled, after closing all processes from Asus it is all OK.
Hello Deodato,
I tried to align by ear the tweeter and mid-woofer with tone generator (for some reason, measuring impulse response with Holm is not giving consistent measurements, maybe the PC is too slow).
Anyway, I noticed that the best sound comes when the tweeter is delayed with -2.7cm, and not with 2.7cm as i would have expected, based on their physical alignment on the baffle. And the question: -2.8 delay for tweeter means that the tweeter fires quicker than the mids? Or it means that tweeter fires later than the mid?
Thank you!
I tried to align by ear the tweeter and mid-woofer with tone generator (for some reason, measuring impulse response with Holm is not giving consistent measurements, maybe the PC is too slow).
Anyway, I noticed that the best sound comes when the tweeter is delayed with -2.7cm, and not with 2.7cm as i would have expected, based on their physical alignment on the baffle. And the question: -2.8 delay for tweeter means that the tweeter fires quicker than the mids? Or it means that tweeter fires later than the mid?
Thank you!
Thank you!
Also want to report that a USB Creative SB 5.1 works better than the Asus, it does not click and pop also. 20Euro...
indeed that sound card is fine piece of electronic,adjustable 48db/octave highpass/mono subwoofer is what made me buy this card ,and it outperformed 4 times priced dac i had,drivers are dream to work with , no bugs.not sure why this card is unpopular.
FIR would be quite possible, but I have no plans in the direction as I do not like the philosophy. As for zero phase, phase distortion is inaudible, as the ear works as a frequency detector, and frequency info and precise timing information for the same wave are mutually exclusive. Thats actually akin to the associativity of moment and location in quantum mechanics, just so zero phase proponents know what they are up against. And the price to pay for zero phase, pre ringing, appears to be quite audible if you ask independent researchers. Furthermore, I am not a fan of DRC, as that always implies optimization for one sweet spot, and deterioration for all others. A speaker sounds best with a near omnidirectional radiation pattern, calibrated in a dead environment. Just let the room(an averagely live one), live. The experience is created by the environment too, and if you like utmost sterility just use a headphone. You can tame that one really obnoxious room mode at say 60 Hz with any parametric eq.
Hi Deodato,
First of all, the ear is not phase deaf. This paper
http://www.bodziosoftware.com.au/Attributes_Of_Linear_Phase_Loudspeakers.pdf
quotes a number of researchers who are arriving at this conclusion (most recent - page 21) and in the process, they emphasize the importance of transients and dealing with time domain distortions.
Take it easy with quantum mechanics - the ear is not trying to do two things at the same time. "The ear works as a frequency detector" - yes, but not only.
It is evident, that the ear examines the incoming audio stimulus in two-stage process:
1. Location – here the transient of the stimulus is examined, and
2. Signal – here the spectral properties of the stimulus are examined.
The two processes always work in-tandem. It is therefore essential, that the loudspeaker provides undistorted waveforms to the auditory system to enable correct processing of both stages.
Localization accuracy is dependent on transients. And not just any transients. Sharp transients with strong attack aid the localization, while slow, smeared transients make localization more difficult. Consequently, the implications for loudspeaker design here is: do not mess up with transients - you must reproduce them as faithfully as possible. So:
Linear phase = transient perfect speaker = undistorted transients = good localization
Second attribute of linear phase loudspeaker is audible in the bass response. The bass remains seismic, deep and powerful, but is also tight and has punch to it now. In short – this bass is accurate.
Square waves are not he only signals that can reveal time-domain (or transient) performance of the loudspeaker. Step function and short pulses are also very useful. On page 2 in this paper: http://www.bodziosoftware.com.au/Fre...Evaluation.pdf
You can see that linear-phase loudspeaker has perfect step response (undistorted transients capability), while traditional loudspeaker is performing quite badly.
Pre-ringing is a dead-horse. This paper examines the issue in details: http://www.bodziosoftware.com.au/Pre_Post_Ringing_IR_And_Pulses.pdf
On the top of this, I have been running large, 5.2HT linear phase system, described on my website Bodzio Software for several years now - no pre-ringing issues whatsoever.
This would be my take on this issue – loudspeakers that are accurate in both: time domain, and in frequency domain.
This can only be done with FIR filters. That's why people are asking for them.
Best Regards,
Bohdan
Hi Julf,
You are entitled to your own opinion. Now would be a good time to prove it.
Best Regards,
Bohdan
Prove what, exactly? Did I misunderstand your tests - are you saying that they were not sighted and subjective?
As to the Aalto research, I still have pretty good contacts there, but as Walpurgis Night / May Day are the biggest party occasion in the academic year up in the Nordic countries, I don't think the guys are reachable until well into next week...
Prove what, exactly? Did I misunderstand your tests - are you saying that they were not sighted and subjective?
As to the Aalto research, I still have pretty good contacts there, but as Walpurgis Night / May Day are the biggest party occasion in the academic year up in the Nordic countries, I don't think the guys are reachable until well into next week...
Hi Julf,
No problem. Let the guys have a good time. I do not have any opportunities to ask them secondary questions, so I can only quote what was presented to AES. The paper is quite interesting.
I went back to the Aalto University paper, which looks mainly at the frequency domain and here is their executive summary:
“Human ability to perceive differences in sounds due to the modification of the phase spectrum is studied in this article. Formal listening tests were arranged using synthetic harmonic complex signals. The tests confirm that humans can perceive differences in the phase spectrum. Furthermore, the perception of phase was found to be somewhat local in frequency, but there is interaction between nearby auditory bands. In addition, the phase spectrum affects the perceived timbre, and the effects are more perceivable the lower the fundamental frequency is. Based on the results, an auditory model for explaining these effects was developed. It aims to mimic the firing rate of the neurons in the cochlea. The output of the auditory model and the listening-test results were compared showing a good match.”
So, looking at the frequency domain alone, you can still discriminate signals with different phase characteristics – thus their general conclusion was, that the “ear in not phase deaf” – I am OK with this conclusion.
The compilation of papers I provided, is aimed to highlight the ongoing research into importance of time domain performance of loudspeakers and also presents conclusions drawn from what is known today. Time domain distortions of signals are attributed/related to phase distortions or phase non-linearity.
Some of the papers look at the phase issues in frequency domain, some papers advocate the importance of typically neglected time domain analysis, some papers consider transients and their effect on localization, and some papers explain the way the ear performs it’s 2-stage analysis.
In all instances, the underlying issue is phase (or time if you like).
What linear-phase system do you use for your listening test?.
Best Regards,
Bohdan
So am I - but the leap from "the ear is not phase deaf" to "linear-phase filters are audibly better" is what I think is a step too far.
Hi Julf,
It looks to me, that we are talking about two different things.
It is not “linear-phase filters” I have described in my poor efforts.
I am advocating linear-phase audio system – the one which has all audio components, most notably loudspeakers themselves, working as linear-phase devices.
I can run my system in linear-phase, or minimum-phase. It takes 2 seconds to switch.
I have described in the papers on me website what I hear when the system runs in linear-phase. The effect on localization is subtle, but is there.
For your information – I like the fact, that loudspeakers are accurate in time domain, I am not fanatical about intricacies of localization (Geddes perhaps may). This is because I mainly use this system for music DVDs and action movies, where surround sound delivers more than enough spatial differentiation for me.
The effect on bass is unmistakable. I do not need to perform blindfolded tests – the effect is too obvious. I have described it many times before.
Once again – what linear-phase system do you listen to?.
Best Regards,
Bohdan
Read this little article, its by scientists from Stanford, not purveyors of a particular filter type https://ccrma.stanford.edu/~jos/filters/Linear_Phase_Really_Ideal.html
Hi Deodato,
Thank you - I know this website.
They have chosen 8-order elliptic-function filter, which frequency response is shown on figure 11.3.
Please take a good look at this filter. Bandpass End = 2000Hz, Stopband begins = 2200Hz. The frequency difference between those two points is only 200Hz, while SPL response difference between those two frequency points is 150dB (yes, 150dB).
I do not know how to translate this characteristics into traditional filter slope ( like12dB/oct for 2nd order LR filter), so I did some simple graphical exercise on the Figure 11.3.
It turns out, that the low-pass slope of the filter they used is well in excess of 1000dB/oct – possible even 1200dBoct or more.
Obviously, this filter belongs to the “brick-wall” category and will ring like hell.
However, I use and recommend LR filters of 12db/oct and 24dB/oct. Their pre-ringing is
80-100 times less and therefore completely inaudible.
In addition, there is a complete cancellation of pre-ringing as described in http://www.bodziosoftware.com.au/Pre_Post_Ringing_IR_And_Pulses.pdf
on page 12 – even on 96dB/oct filters.
Once again – the pre-ringing is a dead horse.
Best Regards,
Bohdan
Hello Bohdan,
The general consensus (brought about by litstening tests..) is still that phase distortion is not audible under normal circumstances. Really, "mimic the firing rate of the neurons in the cochlea"?? Please let blind tests be the judge over this. I will not link to papers by my own hand as evidence. The same inaudibility probably applies to pre echos, under normal circumstances. So if pre ringing is a dead horse, so is ZP, for all audible intents and purposes. But I do find pre echo as a phenomenon repulsive because it is unnatural.
All filters in nature have their phase signature, it is an intrinsic property of filters, electic or mechanic or otherwise. If the frequency response of one filter, or loudspeaker, is compensated by another filter, then the phase response is also compensated.
A perfect compensation cannot be claimed for the ZP filters in a crossover, as the drivers usually have different radiation patterns.
And then there also is the issue of FIR and frequency resolution under 100Hz or thereabouts. Precision, (and no component tolerances etc.) is one of the merits of DSP, but in the case of FIR this precision is a tradeoff with latency; it will take more than 10000 taps to have reasonable precision at 30 Hz. Do the math for your latency and precision. And of course there is the computational overhead for all these taps, so forget about a lean "What You Play Is What You Hear" system with a DSP on a multimedia computer, it will more likely become an audio server kind of solution. Add the price of yer dongle and one might want to look at something like miniDSP gear which I think is kinda cool. But we are talking PC based here of course. And Arcgotic, maybe later I will add FIR as an experiment, if only to assess the claims myself, but not anytime soon I think.
The general consensus (brought about by litstening tests..) is still that phase distortion is not audible under normal circumstances. Really, "mimic the firing rate of the neurons in the cochlea"?? Please let blind tests be the judge over this. I will not link to papers by my own hand as evidence. The same inaudibility probably applies to pre echos, under normal circumstances. So if pre ringing is a dead horse, so is ZP, for all audible intents and purposes. But I do find pre echo as a phenomenon repulsive because it is unnatural.
All filters in nature have their phase signature, it is an intrinsic property of filters, electic or mechanic or otherwise. If the frequency response of one filter, or loudspeaker, is compensated by another filter, then the phase response is also compensated.
A perfect compensation cannot be claimed for the ZP filters in a crossover, as the drivers usually have different radiation patterns.
And then there also is the issue of FIR and frequency resolution under 100Hz or thereabouts. Precision, (and no component tolerances etc.) is one of the merits of DSP, but in the case of FIR this precision is a tradeoff with latency; it will take more than 10000 taps to have reasonable precision at 30 Hz. Do the math for your latency and precision. And of course there is the computational overhead for all these taps, so forget about a lean "What You Play Is What You Hear" system with a DSP on a multimedia computer, it will more likely become an audio server kind of solution. Add the price of yer dongle and one might want to look at something like miniDSP gear which I think is kinda cool. But we are talking PC based here of course. And Arcgotic, maybe later I will add FIR as an experiment, if only to assess the claims myself, but not anytime soon I think.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.