Is it necessary to AUTOMATICALLY change the convolution filters in accordance with the changes in the sample rate of the reproduced material?

[SashaR]

So have a go, try to understand the basic concepts.

That's great! You have described my current situation absolutely accurately!!!
The problem is that I cannot occupy myself with studying this science 24*7 - for various reasons . Therefore, I am looking for an opportunity to follow in someone's footsteps by choosing someone who claims that he has achieved the result and it is good. I have already realized that in order to perform the procedure as a robot, I will still have to master the language (terms) and TECHNICAL (THOUGHTLESS) EXECUTION OF THE COMMANDS RECORDED ON IT.

I would REALLY appreciate your opinion on two issues!
The first: in addition to reducing the amount of calculations during DSP operation, what does replacing my current PEQ with a convolution filter (even in its current state, it already gives a 1 dB deviation from the target from 20 to 17k)? This PEQ currently consists of 39 filters, and I hope that it may not stutter on the PC that I intend to use for DSP (i3).

The second:
is (based on your feelings from what you read in the next thread - mainly in the last post of this thread) - is the procedure described in it capable of creating a GOOD result (at least superior in sound quality to what the PEQ that I have now has done: it even leveled a hole with a depth of 24 and 21 dB below the target level (its depths are 6 Hz apart!) and the deviations on the entire target do not exceed 1 dB, as I already told you), and
B) can Dummy be able to perform the described procedure if he learns how to implement the operations described in this procedure? I am asking this question on the assumption that anyone can take my measurement file and perform on their machine all the necessary steps of the procedure described here if they know HOW to perform these ACTIONS (even if they do not understand their inner essence!).

FIR filters via REW?

Hello, Can anybody help how to make FIR filterset for headphones? I'm familiar with taking measurements and autoeq in REW, but I don't know how to start with FIR fitlers. Is it possible to make it with rephase? Is there any android app to apply them later? I have Minidsp EARS and Focusrite 2i2...

www.audiosciencereview.com

Thank you very much in advance for your help!!!

 

[SashaR]

I mastered, whose in touch and participated in studis when I was a bit younger. Nothing to do with sample rate but sample time (SINAD) and analytical listening on rather loud and above average SPL and even so subtle 24 bit to 16 bit without dithering and with a good one not even that. Difference between master's on other hand is very, very real! Good, great and broken. Great is when something extraordinary happens during recording and it simple turns out like that, good and technically done excellent even with gear shortcomings and pretty much neither done good or correct and simply broken (and that's a lot of music).

NO, NOT REALLY! I was talking, for example, about a set of files made from a single real master (tape or file), each of which is encoded in different formats (PCM, MQA, DSD) with different sampling rates. The difference is audible.
In addition, there is a noticeable difference in the different versions of the same album that TIDAL offers (44, 96, 192).

That is, I was talking only about the technical, not the artistic difference, which is audible - audible despite the theory, which does not take into account the mass of what is not yet known to this theory. And the brain hears this "something"

With great respect (and envy for your knowledge of the technique of creating convolution filters)

 

[Keith_W]

That's great! You have described my current situation absolutely accurately!!!
The problem is that I cannot occupy myself with studying this science 24*7 - for various reasons . Therefore, I am looking for an opportunity to follow in someone's footsteps by choosing someone who claims that he has achieved the result and it is good. I have already realized that in order to perform the procedure as a robot, I will still have to master the language (terms) and TECHNICAL (THOUGHTLESS) EXECUTION OF THE COMMANDS RECORDED ON IT.

I would REALLY appreciate your opinion on two issues!
The first: in addition to reducing the amount of calculations during DSP operation, what does replacing my current PEQ with a convolution filter (even in its current state, it already gives a 1 dB deviation from the target from 20 to 17k)? This PEQ currently consists of 39 filters, and I hope that it may not stutter on the PC that I intend to use for DSP (i3).

The second:
is (based on your feelings from what you read in the next thread - mainly in the last post of this thread) - is the procedure described in it capable of creating a GOOD result (at least superior in sound quality to what the PEQ that I have now has done: it even leveled a hole with a depth of 24 and 21 dB below the target level (its depths are 6 Hz apart!) and the deviations on the entire target do not exceed 1 dB, as I already told you), and
B) can Dummy be able to perform the described procedure if he learns how to implement the operations described in this procedure? I am asking this question on the assumption that anyone can take my measurement file and perform on their machine all the necessary steps of the procedure described here if they know HOW to perform these ACTIONS (even if they do not understand their inner essence!).

FIR filters via REW?

Hello, Can anybody help how to make FIR filterset for headphones? I'm familiar with taking measurements and autoeq in REW, but I don't know how to start with FIR fitlers. Is it possible to make it with rephase? Is there any android app to apply them later? I have Minidsp EARS and Focusrite 2i2...

www.audiosciencereview.com

Thank you very much in advance for your help!!!

I do not recommend that a dummy mechanically follow instructions on how to produce a FIR filter with REW. With manual filter generation, you have to look at the measurement and make decisions. And there are A LOT of decisions to make. I can understand that learning all this stuff is not for everyone, but unfortunately for you, if you don't want to have to do it manually, you will have to pay some money. Your options are:

- software that holds your hand and makes decisions for you. Example: Dirac.
- software that helps you along, but still requires you to make some decisions. Example: Audiolense.
- pay someone like Mitch (from Accurate Sound) or Thierry (from Home Audio Fidelity) to make filters for you.

I did say in the other thread that in MY opinion, a manually corrected linear-phase filter sounds better. But that's just my opinion. Lots of people disagree, and that's fine.

FWIW I am in the middle of developing a procedure for generating FIR filters with REW. So far I have written 18 pages, and when it's finished, I think i'll be pushing 30-40 pages. Simply because things which take one step with other software takes a few steps in REW.

 

[ZolaIII]

For such people FIR will stay starship and FP a Sci-fi until one day they get abducted (by aliens) and guess what, no one notices it.

 

[SashaR]

I do not recommend that a dummy mechanically follow instructions on how to produce a FIR filter with REW. With manual filter generation, you have to look at the measurement and make decisions. And there are A LOT of decisions to make. I can understand that learning all this stuff is not for everyone, but unfortunately for you, if you don't want to have to do it manually, you will have to pay some money. Your options are:

- software that holds your hand and makes decisions for you. Example: Dirac.
- software that helps you along, but still requires you to make some decisions. Example: Audiolense.
- pay someone like Mitch (from Accurate Sound) or Thierry (from Home Audio Fidelity) to make filters for you.

I did say in the other thread that in MY opinion, a manually corrected linear-phase filter sounds better. But that's just my opinion. Lots of people disagree, and that's fine.

FWIW I am in the middle of developing a procedure for generating FIR filters with REW. So far I have written 18 pages, and when it's finished, I think i'll be pushing 30-40 pages. Simply because things which take one step with other software takes a few steps in REW.

Thanks a lot for the reply! I wish you good luck in your work on creating the tutorial! I hope that he is, after all! - it will not be an academic course, the FULL CONTENT of WHICH the student must remember, but will allow people who do not have the opportunity (or the necessary level of pride) to master ALL this science to receive help and support in places where your work will help a person to realize at the most minimal level the essence of the problem, because of which it is necessary to make a decision about the direction of further movement. A book that is a step-by-step guide that does not try to bring down on a person who wants to go to the store all the information about underground communications located under the sidewalk along which a person is escorted to the store. It is probably undesirable to tell the traffic light switching scheme and the exact timing of their interaction with each other and with the environment at street intersections...

A book that is a guide, not a textbook... A SIMPLE guide, with constant help provided as much as possible!


I would appreciate your answer to the question I asked you about the procedure described in the link I sent.
And I would appreciate your developing the idea of paying for the creation of the filter(s) - I need three POINT filters for my apartment: which of the filter purchase options (tools, specialists) will be cheaper and better (in terms of sound quality)? (And easier...)

By the way, based on your experience using filters, can you tell me what the DSP implemented in Windows10 on the i3 processor with a frequency of ~1.9-~3.0 MHz (if I'm not mistaken) will be capable of. Will it be able to support my PEQ of 39 filters? Will it be able to support a convolution filter made for the same system as this PEQ? Any advice and opinions from you will be greatly appreciated. (I have a simple system of two 30-35k speakers - and that's it! And a tiny room with V=32 m^3).

Thank you again for your help!

 

[boxerfan88]

Tell me, please this tweak!

Ok, I found my old notes.
Here it is change it from 0xFFFFFFFF to 0xa (10 decimal), then reboot PC.
Warning - this will sacrifice network traffic throughput.

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Multimedia\SystemProfile\NetworkThrottlingIndex
    Off: DWORD Value 0xFFFFFFFF (Hex)
    On: DWORD Value 10 (Decimal), Default - Range: Decimal 1-70

I would also like a way to increase the priority of the EAPO DSP, but I can't find
it in the list of processes...

My suggestion is don't bother ... Windows OS should take care of APO priority.
If you really wanna control process priority ... try the software called Process Lasso. There is a free version (limited functionality).
I was using Process Lasso when I was experimenting with VSThosts ... but I managed to get all my DSP inside foobar therefore I don't need VSThosts nor Process Lasso anymore.
URL: https://bitsum.com/

.

 

[Keith_W]

I would appreciate your answer to the question I asked you about the procedure described in the link I sent.

Sorry, that would involve me having to watch that 30 minute video, writing notes, and then commenting on it. I am not going to do that. If you like, you can write your own procedure based on that video and submit it to ASR. We can comment on it then.

 

[SashaR]

Thank you very much for your help and advice!!!


P.S. Please look at the procedure described in this post: is it the right decision to choose this procedure for constructing a convolution filter? My absolute incompetence has decreased a little in the last few days - and so now I know that the dips in the frequency response that bother me most are located before the transition zone, before the Schroder frequency. Can this particular procedure help me elegantly deal with the 24 dB dip in the frequency response, without -20 dB preamplification?

https://audiosciencereview.com/forum/index.php?threads/fir-filters-via-rew.52819/post-1967369

 

[SashaR]

Sorry, that would involve me having to watch that 30 minute video, writing notes, and then commenting on it. I am not going to do that. If you like, you can write your own procedure based on that video and submit it to ASR. We can comment on it then.

I ask you to excuse me for my mistake: I was not referring to the first post in this thread, but to this one, you yourself participated in this conversation!
Can this particular procedure help me elegantly deal with the 24 dB dip in the frequency response, without -20 dB preamplification?

FIR filters via REW?

Hello, Can anybody help how to make FIR filterset for headphones? I'm familiar with taking measurements and autoeq in REW, but I don't know how to start with FIR fitlers. Is it possible to make it with rephase? Is there any android app to apply them later? I have Minidsp EARS and Focusrite 2i2...

audiosciencereview.com

Please forgive me for the wrong link I sent!

 

[ZolaIII]

Thank you very much for your help and advice!!!


P.S. Please look at the procedure described in this post: is it the right decision to choose this procedure for constructing a convolution filter? My absolute incompetence has decreased a little in the last few days - and so now I know that the dips in the frequency response that bother me most are located before the transition zone, before the Schroder frequency. Can this particular procedure help me elegantly deal with the 24 dB dip in the frequency response, without -20 dB preamplification?

https://audiosciencereview.com/forum/index.php?threads/fir-filters-via-rew.52819/post-1967369

He vogue describes how to do impulse response negative and add to it PEQ along with crossover and sum it up (impuls negative +PEQ filter export). And do all pass filter at the end. Nothing will correct 24 dB deep. Try by placement to improve it to be either smaller or narrow. Then do the EQ-ing, you can add digital gain (in FP again) to match analog one when you for various reasons lose it (normalisation and such). When you EQ first by hand kick the room fundamental frequency down (single PEQ peek filter) to the level of average refractions around it in spectral plots. With it in place do impuls negative, put it on convolver and do measurements with it. Let REW make PEQ for it, generate filter graph from it. Sum impuls negative with generated filter graph and bake convolution out of it and use it as final one. In the end play with phase correction and all pass linear phase (which you can also merge into convolution filter same way). If it doesn't sound clear you must start working and after some time doing it it will came up to you. Remember don't try to correct deeps more than +2 dB in either impulse negative or while generating PEQ! I already gave you video which is good for impulse alignment and inversion (skip the VBA that's what that single PEQ by hand does) and to figure out how operations are done in REW. Rest is more of the same and playing a lot with the phase.

 

[Keith_W]

I ask you to excuse me for my mistake: I was not referring to the first post in this thread, but to this one, you yourself participated in this conversation!
Can this particular procedure help me elegantly deal with the 24 dB dip in the frequency response, without -20 dB preamplification?

The answer is, nobody knows without looking at the measurement! You can not simply take a dip and invert it. Sometimes you can, but if your dip is 24dB dip it is very likely that you can't. Dips are unlikely to be minimum phase, and therefore probably can't be corrected by inversion. The solution isn't always DSP, sometimes it's moving speakers or subwoofers, or buying more subwoofers.

 

[SashaR]

He vogue describes how to do impulse response negative and add to it PEQ along with crossover and sum it up (impuls negative +PEQ filter export). And do all pass filter at the end. Nothing will correct 24 dB deep. Try by placement to improve it to be either smaller or narrow. Then do the EQ-ing, you can add digital gain (in FP again) to match analog one when you for various reasons lose it (normalisation and such). When you EQ first by hand kick the room fundamental frequency down (single PEQ peek filter) to the level of average refractions around it in spectral plots. With it in place do impuls negative, put it on convolver and do measurements with it. Let REW make PEQ for it, generate filter graph from it. Sum impuls negative with generated filter graph and bake convolution out of it and use it as final one. In the end play with phase correction and all pass linear phase (which you can also merge into convolution filter same way). If it doesn't sound clear you must start working and after some time doing it it will came up to you. Remember don't try to correct deeps more than +2 dB in either impulse negative or while generating PEQ! I already gave you video which is good for impulse alignment and inversion (skip the VBA that's what that single PEQ by hand does) and to figure out how operations are done in REW. Rest is more of the same and playing a lot with the phase.

Thank you very much for the instructions!!! I'll try to start reading it by studying the words... Thank you!

Remember don't try to correct deeps more than +2 dB in either impulse negative or while generating PEQ!

One question before I start learning the words: REW offers everyone the opportunity to make a PEQ that will provide a 1 dB deviation from the target - do not use its offer? Or is it an instruction that relates to the process you described and does not relate to the single creation of a PEQ?

Thank you!!!

 

[SashaR]

Dips are unlikely to be minimum phase

The -24 and -21 dB dips that I have are exactly in these two modes: 112 and 119 Hz. They are the inherent "part" of the room, unfortunately

REW built me a PEQ that turned the frequency response into a straight line - but it costs -20 dB of pre-amplification of the entire range (20-17k)! Now I'm trying to use EAPO as a DSP, and it turns out that it has the ability to use scripts... Does such a beast as spot, local ("spotted") pre-amplification exist in nature? Or limiting the level of the resulting signal only in a selected part of the range?

 

[ZolaIII]

Thank you very much for the instructions!!! I'll try to start reading it by studying the words... Thank you!

One question before I start learning the words: REW offers everyone the opportunity to make a PEQ that will provide a 1 dB deviation from the target - do not use its offer? Or is it an instruction that relates to the process you described and does not relate to the single creation of a PEQ?

Thank you!!!

It's like conditioner for him to try to stay in to the given boundaries, not meaning it will turn out that good just a target. It's important to adresa the room fundamental first (along with it's two harmonics, three peeks and Two deeps related to it) to get it out of equation. Then you let REW do it's auto generating part. If you do impuls time alignment and inversion, you apply them and then let the REW auto PEQ measurements of it, export it as measurements (PEQ) and sum A impuls inversion +B PEQ measurements to make new convolution out of the output of it.

 

[ZolaIII]

The -24 and -21 dB dips that I have are exactly in these two modes: 112 and 119 Hz. They are the inherent "part" of the room, unfortunately

REW built me a PEQ that turned the frequency response into a straight line - but it costs -20 dB of pre-amplification of the entire range (20-17k)! Now I'm trying to use EAPO as a DSP, and it turns out that it has the ability to use scripts... Does such a beast as spot, local ("spotted") pre-amplification exist in nature? Or limiting the level of the resulting signal only in a selected part of the range?

View attachment 447034

Obviously it's about 4.5m long room so 39, 78, 117 Hz and 47, 67 Hz deeps... Ufff read the post above.

 

[boxerfan88]

The -24 and -21 dB dips that I have are exactly in these two modes: 112 and 119 Hz. They are the inherent "part" of the room, unfortunately

You should ignore all those gigantic nulls. Those are due to cancellation at the measurement location. No matter how much gain you throw at it, most of the gain will cancel (ie ineffective in solving the gigantic dips).

IME those gigantic narrow dips are not that audible. My guess is that our ears-brain system can hear through those narrow dips.

 

[SashaR]

You should ignore all those gigantic nulls

How to do this ignoring in REW - I mean now only regular PEQ, not convolution.

After I get my microphone back I will test your assumption about the impossibility of filling the bottomless barrel of modality

 

[boxerfan88]

How to do this ignoring in REW - I mean now only regular PEQ, not convolution.

Set the individual max boost to 0.

 

[SashaR]

Set the individual max boost to 0.

View attachment 447118

The range must be so narrow (30-300 only?)? What about the rest (30-17k)? Make it separetdly?

 

[boxerfan88]

The range must be so narrow (30-300 only?)? What about the rest (30-17k)? Make it separetdly?

Oh, ignore that. Please use whatever correction range you prefer.

I have my own philosophy about room correction that may not apply to other people. I normally let REW run wild (do whatever it wants) from 30-300 only. I use wide-Q method to correct between 300-1000, and normally restrict myself to 1-2 PEQ in this area. I do not correct above 1kHz.