DSP Xover project (part 2)

Jump to Latest
#943
JLOP said:
Hi all

I discussed this week-end with a great specialist of filters (I wont name as I am not sure I translate properly 100% his advice).
Basically, he told me
- below #200 Hz, pre-ringing from FIR maybe heard, so best to limit FIR above that threshold, most of the pre-ringing will disappear
- phase changes is mostly heard between #200 and #4000Hz, so it is nice to that phase is corrected in this range

Will this be compatible with the mixed FIR/IIR approach that has been discussed ?

BR
Jean-Louis
Click to expand...

This would be no problem at all. I don't agree with the statements, though.
 
#945
Recently I have played two setups against each other:
1. Correction on the mid horn JBL2482 compression drivers
2. No correction.

In A/B test and blind tests (helper changing the setups with me having no idea what the starting point is), I clearly prefer them without any correction...
They sound is just less appealing and natural.

Now, I assume that the correction is pushing the the driver / horn combo to do work where out of their comfort zone and so it sounds less good.

Is it the way I am applying correction or just, you can't get things for nothing?

I do however like some correction on the bass channels. Perhaps it is a compression driver thing? Same thing with my upper mid Vitavox S2 drivers btw...
 
#946
JLOP said:
Hi Kazam

Thanks for the answer about feasibility.
For my understanding, what are you disagreeing with specifically ?

BR
Click to expand...

I don't want to make this into a Linear Phase thread but very briefly here are the four main reasons why linear phase is not important in audio proessing:

  1. Because listening tests have shown that it is not. See "Sound reproduction" by Floyd Toole, pages 419-421.
  2. Because if you take the opposing view and say that linear phase is indeed important what would that mean? The human auditory system evolved to maximize speech recognition. The spectral content is roughly 1-3kHz. Obviously the wavelength is 3 times shorter at 3kHz as compared to 1kHz. So phase traverses 3 times as fast in space when the sound source moves. If phase coherence was important that would mean that you would think that someones voice changed considerably as they moved through a room. But obviously that doesn't happen so the initial assumption must be wrong.
  3. Microphones were used to capture the recording you are listening to. Microphones are not linear phase so even if your loudspeaker is, it doesn't matter. It doesn't mean what you get is more "true" to the event.
  4. Because even if you don't buy the above reasons there is a simple test you can do yourself. Simply generate a sum of sine signals with different phases and see if you can hear the difference. You will find that even though the waveforms are readily distinguishable from each other on a time plot they sound identical. I have done this test myself. Quite educational.

Now, rapidly changing phase introduces ringing in the impulse response so I'm not saying phase doesn't matter. It does. It just doesn't have to be linear.

/K
 
#947
kazam said:
I don't want to make this into a Linear Phase thread but very briefly here are the four main reasons why linear phase is not important in audio proessing:

  1. Because listening tests have shown that it is not. See "Sound reproduction" by Floyd Toole, pages 419-421.
  2. Because if you take the opposing view and say that linear phase is indeed important what would that mean? The human auditory system evolved to maximize speech recognition. The spectral content is roughly 1-3kHz. Obviously the wavelength is 3 times shorter at 3kHz as compared to 1kHz. So phase traverses 3 times as fast in space when the sound source moves. If phase coherence was important that would mean that you would think that someones voice changed considerably as they moved through a room. But obviously that doesn't happen so the initial assumption must be wrong.
  3. Microphones were used to capture the recording you are listening to. Microphones are not linear phase so even if your loudspeaker is, it doesn't matter. It doesn't mean what you get is more "true" to the event.
  4. Because even if you don't buy the above reasons there is a simple test you can do yourself. Simply generate a sum of sine signals with different phases and see if you can hear the difference. You will find that even though the waveforms are readily distinguishable from each other on a time plot they sound identical. I have done this test myself. Quite educational.

Now, rapidly changing phase introduces ringing in the impulse response so I'm not saying phase doesn't matter. It does. It just doesn't have to be linear.

/K
Click to expand...


Hi Kazam,

Regarding the importance of phase in audio, please have a look at the paper:

http://www.bodziosoftware.com.au/Attributes_Of_Linear_Phase_Loudspeakers.pdf



Best Regards,
Bohdan
 
#951
Regarding remote controls...

In the US almost all remotes that are supplied by cable television providers can be programmed to operate nearly every brand of TV, DVD, stereo equipment, etc. I tested mine with my "office" JVC receiver and it worked. Free, and the batteries last a long time. Perhaps someone can test Najda.
 
#952
kazam said:
I have read it and disagree with the conclusions.

I suggest we discuss issues related to the Najda board in this thread. Like I said, I don't want to turn this into a linear phase thread.
Click to expand...


Hi Kazam,

Well, you started the "linear phase" issue in this thread.

Anyway, you are welcomed to have a different opinion, but unless you present something more technically minded, your conclusion remains just an opinion.

Best Regards,
Bohdan
 
#953
JLOP said:
Very interesting summary paper, thanks Bohdan. I guess it must be quite easy to A/B check with Najda with and without phase correction ?

Kind regards
Click to expand...


Hi Jlop,

Thank you for your kind comments.

Regarding Najda, unfortunately I do not have any experience with this product. From what I read on this thread, this DSP is a "low processing power" device, that can manage just 1023 taps/channel.

I have been operating Ultimate Equalizer based 5.2HT system for almost 9 months now ( described on Bodzio Software ). Ultimate Equalizer V4 uses 16384 taps for linear-phase filters with short latency, or 32768 taps for further improved low-frequency resolution at 48kHz or 96kHz.

It does this for 8in/16out channel configuration (used for HT applications), so each channel can be processed with 32768 FIR taps. You will need this (not IIR) for processing phase. As you can see, these are all identical channels with absolutely no restrictions.

The whole system is SPL and phase equalized (both to flat line, 0deg phase).
I no longer listen to minimum-phase audio systems. You will find a lot of information on the above website.

Let me know if I can answer any other of your questions.

Best Regards,
Bohdan
 
#956
bohdan1232000 said:
Hi Kazam,

Well, you started the "linear phase" issue in this thread.

Anyway, you are welcomed to have a different opinion, but unless you present something more technically minded, your conclusion remains just an opinion.

Best Regards,
Bohdan
Click to expand...

From "Sound Reproduction", Floyd Toole:

Toole (1986) shows phase responses for 23 loudspeakers arranged according to subjective preference ratings. The most obvious relationship was that those with the highest ratings had the smoothest curves, but linearity did not appear to be a factor. The agreement that smoothness is desirable argues that listeners were attracted to loudspeakers with minimal evidence of resonances because resonances show themselves as bumps in frequency response curves and rapid up-down deviations in phase response curves. The most desirable frequency responses were also horizontal straight lines. The corresponding phase responses had no special shape other than the smoothness. This suggests that we like flat amplitude spectra and we don’t like resonances, but we tolerate general phase shift, meaning that waveform fidelity is not a requirement.
Click to expand...

My emphasis.

The AES paper referenced is:
“Loudspeaker Measurements and Their Relationship to Listener Preferences,”
J. Audio Eng. Soc., 34, pt. 1, pp. 227–235; pt. 2, pp. 323–348.

The issue has also been investigated by Siegfried Linkwitz in a general call to arms: Phase distorsion effects
The conclusion is that phase distorsion is not audible, if kept within reasonable limits, nor a requirement for hi-fidelity sound reproduction.

As I mentioned I have also convinced myself by listening to various test tones and program segments and found no evidence of the superiority of phase linear filtering. Long linear-phase FIR filters introduce a lot of pre-echo due to their symmetric impulse response which is a very undesirable and unnatural effect. It should be avoided.

I stand by my original statement and also add that it is more than "just an opinion" as the above references should make clear.

/Kazam
 
#957
bohdan1232000 said:
Regarding Najda, unfortunately I do not have any experience with this product.
Click to expand...

Then why are you posting in this thread?

I have been operating Ultimate Equalizer based 5.2HT system for almost 9 months now ( described on Bodzio Software ). Ultimate Equalizer V4 uses 16384 taps for linear-phase filters with short latency, or 32768 taps for further improved low-frequency resolution at 48kHz or 96kHz.
Click to expand...

Aha, you have something to sell.

It does this for 8in/16out channel configuration (used for HT applications), so each channel can be processed with 32768 FIR taps. You will need this (not IIR) for processing phase. As you can see, these are all identical channels with absolutely no restrictions.
Click to expand...

This seems rather limited. BruteFIR claims:
With a massive convolution configuration file setting up BruteFIR to run 26 filters, each 131072 taps long, each connected to its own input and output (that is 26 inputs and outputs), meaning a total of 3407872 filter taps, a 1 GHz AMD Athlon with 266 MHz DDR RAM gets about 90% processor load, and can successfully run it in real time. The sample rate was 44.1 kHz, BruteFIR was compiled with 32 bit floating point precision, and the I/O delay was set to 375 ms. The sound card used was an RME Audio Hammerfall.
Click to expand...
It is also free.

/Kazam
 
#958
kazam said:
Then why are you posting in this thread?



Aha, you have something to sell.



This seems rather limited. BruteFIR claims:
It is also free.

/Kazam
Click to expand...


Hi Kazam,

Last time I checked (it was a couple of years ago, I admit) BruteFIR was just a convolution engine. It does not do anything else. Latency of 375ms is high, but perhaps typical for this type of FIR lengths.

UE4 is a very sophisticated HBT equalizer / crossover / playback system / measurement system and runs two sound cards in synch.

No comparison there.

Best Regards,
Bohdan
 
#959
kazam said:
From "Sound Reproduction", Floyd Toole:



My emphasis.

The AES paper referenced is:
“Loudspeaker Measurements and Their Relationship to Listener Preferences,”
J. Audio Eng. Soc., 34, pt. 1, pp. 227–235; pt. 2, pp. 323–348.

The issue has also been investigated by Siegfried Linkwitz in a general call to arms: Phase distorsion effects
The conclusion is that phase distorsion is not audible, if kept within reasonable limits, nor a requirement for hi-fidelity sound reproduction.

As I mentioned I have also convinced myself by listening to various test tones and program segments and found no evidence of the superiority of phase linear filtering. Long linear-phase FIR filters introduce a lot of pre-echo due to their symmetric impulse response which is a very undesirable and unnatural effect. It should be avoided.

I stand by my original statement and also add that it is more than "just an opinion" as the above references should make clear.

/Kazam
Click to expand...



Hi Kazam,

If you do not have a phase-linear system – do not worry. The sky will not fall down.
The differences are not earth-shattering by any means.

There are basically three areas where linear-phase loudspeakers differ from. minimum-phase loudspeakers.

  • Linear-phase speakers provide more accurate spatial information, rather than timbral. Tonal balance is the same for both loudspeaker types. This is where the tests are falling apart, because listeners are looking for tonal differences, rather than subtle spatial clues – sharper image, better located soloists, stage depth.
  • Identical phase response for all loudspeaker in the system. The phase response in correctly equalized multi-channel linear-phase system is 0deg in every loudspeaker. Therefore it immediately satisfies AESTD1001.1.01-10 to perfection.
  • Tighter bass. Even Floyd Toole quoted other researchers (Craven and Gerzon) on this subject on page 420.
Anyway, I can quote all research results I can find, but at the end of the day, if I could not hear or measure these differences, then I would not care about the linear-phase at all.

So, measurements of linear-phase loudspeaker are presented on my website.
I (and other sources) can testify to more accurate spatial information. It’s subtle, but it is there.

The most obvious difference is the tighter bass. I have conducted extensive tests on this subject - http://www.bodziosoftware.com.au/LP_MP_Subwoofer_Tests.pdf
You simply can not miss it.

Impulse response issues are examined in this paper.
http://www.bodziosoftware.com.au/Pre_Post_Ringing_IR_And_Pulses.pdf

As far as your own tests are concerned – your description is indicative, that you have no idea about linear-phase manifestations in loudspeakers. Good luck.

Best Regards,
Bohdan
 
You must log in or register to reply here.
Top Bottom