Erwin,
Sorry for my mistype. 4 out and 2 in is correct. I use foo_dsp_xover currently but looking for a good cheap software that does crossover/equalization. The Asus xonar D2 has slightly better S/N than the Audiophile 192 and can do 4 way cross. I use the 192 for testing because it handles high voltage better and I have destroyed a few sound cards in the past with foolish blunders while testing.
Do not get excited about what sound cards are recommended. The software authors just put down what they have used and do not list every card that would work. I am interested in ultimate equalizer but price is a little high for me.
Ted
Sorry for my mistype. 4 out and 2 in is correct. I use foo_dsp_xover currently but looking for a good cheap software that does crossover/equalization. The Asus xonar D2 has slightly better S/N than the Audiophile 192 and can do 4 way cross. I use the 192 for testing because it handles high voltage better and I have destroyed a few sound cards in the past with foolish blunders while testing.
Do not get excited about what sound cards are recommended. The software authors just put down what they have used and do not list every card that would work. I am interested in ultimate equalizer but price is a little high for me.
Ted
Hi Pascal,
I am not sure how much you would value my opinion, since I designed the Ultimate Equalizer, therefore you would be correct suspecting, that I have vested interests in promoting it. However, many people on this board are doing it, therefore I decided to chime in.
Firstly, I would refer you to posts #22, #24 and #30 in this thread. These are all very valued comments, as the come from independent users of the UE.
I have been operating my 5.2 Linear-Phase HT system for over 8 months now. My initial impressions are summarized in http://www.bodziosoftware.com.au/Home_Theatre_Conclusions.pdf page 3-8.
Please also read comment 6 and 7 on page 8/9. I have specifically used vented enclosures for my speakers, as they allow for low-frequency equalization to extend the F3dB cut-off by up to an octave down. All this happens without exceeding linear cone excursions. My front/surround loudspeakers are all quite modest in size, but after HBT equalization and phase linearization, they have surprisingly low and well controlled bass.
You will find quite a lot of information on my website. Amongst others, I would particularly recommend http://www.bodziosoftware.com.au/Attributes_Of_Linear_Phase_Loudspeakers.pdf
paper.
8 months down the track, I continually find the system (PC audio server + UE4) to be the best investment I have made. Let me know if you have more questions.
Best Regards,
Bohdan
I am not sure how much you would value my opinion, since I designed the Ultimate Equalizer, therefore you would be correct suspecting, that I have vested interests in promoting it. However, many people on this board are doing it, therefore I decided to chime in.
Firstly, I would refer you to posts #22, #24 and #30 in this thread. These are all very valued comments, as the come from independent users of the UE.
I have been operating my 5.2 Linear-Phase HT system for over 8 months now. My initial impressions are summarized in http://www.bodziosoftware.com.au/Home_Theatre_Conclusions.pdf page 3-8.
Please also read comment 6 and 7 on page 8/9. I have specifically used vented enclosures for my speakers, as they allow for low-frequency equalization to extend the F3dB cut-off by up to an octave down. All this happens without exceeding linear cone excursions. My front/surround loudspeakers are all quite modest in size, but after HBT equalization and phase linearization, they have surprisingly low and well controlled bass.
You will find quite a lot of information on my website. Amongst others, I would particularly recommend http://www.bodziosoftware.com.au/Attributes_Of_Linear_Phase_Loudspeakers.pdf
paper.
8 months down the track, I continually find the system (PC audio server + UE4) to be the best investment I have made. Let me know if you have more questions.
Best Regards,
Bohdan
The concept of phase equalization creates a better spacial image perception, which is mostly more realistic depth perception. To obtain best results depends on how you measure and the equipment equipment you use. Basically, you want to use as much of your playback electronics as possible; my personal experience is that near field measurements provide the better results.
Having experienced the latest version, it has become an indispensable part of my system. Currently I use it on a full range driver system. In the last few versions, I had not used it because it needed to be the player, but with the latest version, you can design the filter and export a convolution wav file, or export for use with some minidsp models.
I currently use Jriver media center with dsp convolution, the sound becomes shockingly realistic to me.
I currently use Jriver media center with dsp convolution, the sound becomes shockingly realistic to me.
Hello Soonsc,
Does the option of exporting the convulution file into JRiver als mean the hardware on which UE runs needs to be less heavy?
From the UE documentation I get the impression nothing short of a very fast Core i7 machine is needed, but I may be mistaken here. Maybe Bohdan can also comment.
I would like to know what the minimum requirement would be to run UE.
Thanx and regards,
Eelco
Does the option of exporting the convulution file into JRiver als mean the hardware on which UE runs needs to be less heavy?
From the UE documentation I get the impression nothing short of a very fast Core i7 machine is needed, but I may be mistaken here. Maybe Bohdan can also comment.
I would like to know what the minimum requirement would be to run UE.
Thanx and regards,
Eelco
Hi Boden,
When I first started using UE systems for 16-output HT surround sound system four years ago, the CPU that was able to handle the computational load within the required time window was i7-960 device, with appropriate motherboard.
I have built a number of UE systems, and I have always listed the type of motherboard I used. This is because if anybody wanted to duplicate the UE system, it would guarantee that the final outcome would by fully working UE system.
Obviously, if you intend to run 2x4-way stereo system, the CPU requirements would be way down.
The i7-960 is an old CPU by todays’ standards and is benchmarked by https://cpubenchmark.net/cpu.php?cpu=Intel+Core+i7-960+%40+3.20GHz&id=838
at 5924 level.
Contemporary CPUs are much better and for example i7-6700 http://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i7-6700+%40+3.40GHz
Is benchmarked at 9922 level.
Another good CPU, the i4770k https://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i7-4770K+%40+3.50GHz
Is benchmarked at 10169 level.
As you can see, the performance level goes up almost every year, so it means, that much less expensive contemporary CPU will handle the 16-channel workload.
But the above observations are only valid when you actually intend to run UE9 on your own PC, and would include the rest of the motherboard performance.
UE9 exports impulse responses in WAV files, and in ASCII files compatible with miniDSP boards, so now users started to use UE9 with jRiver or any other convolver, as the preferred convolver and also free-standing DSP boards.
This is great, and the CPU demands are those that are required by the jRiver to run and not by UE own playback system.
The length of the exported impulse responses are specified on page 36 in the manual http://www.bodziosoftware.com.au/UE9_Manual.pdf
So you can determine if jRIver could play such files and what CPU requirements are needed.
George Soong has done some work with jRiver, and could probably answer some more detailed questions.
Please let me know if you have any other questions.
Best Regards,
Bohdan
Currently I had a two channel full range driver setup. I have Jriver running the convolution on a Late 2008 13" MacBook aluminum unibody using a Meridian Explorer DAC. Filter design is on a Late 2011 MacBook Pro running Windows from a Parallels virtual machine. I have not explored multi way configurations yet, but for pure music playback, I think it the computer would not be a problem with simple systems. I am looking to explore Xonar U7 under Windows and OSX. But for some reason mine now can only play 16bit 48KHz on all the machines I have, have not found out why it is degraded over time. So I may be getting the Echelon edition to also try out. My goal is to find more affordable entry level high performance options.
Thanks arcgotic, I have been trying many things including that driver. However, the problem seems whenever I plug it into a computer now, Windows always tells me the device can run faster when plugged into a faster port. And it only happens with this device. This did not happen when I first started to use this. I have tried two computers, and the results were the same. It seems just weird, like a person raging, and gradually slowing down. I had it as a sound reference for a while running on an Asus A3500N. When it started to go weird, I got the second hand Late 2008 MacBook to run BootCamped Windows 7. It also tells me that the device can run faster on a faster port.
Anyway, I am hoping to run at least four channels using the the convolving filter generated by UE. The next test is to run filters of different sample rates because Jriver allows automatic switching of the filters of different sample rates. Currently I use the 96KHs sample rates with the largest buffer size. One advantage of using Jriver for this is that you can concurrently setup different zone names to the same output sound card and switch between the different filters to see if you hear a difference or not. This was a fun experience for me because I can choose between no filter, minimum phase equalisation, phase linearisation only, and the full equalisation.
One occurrence when I just moved to the 2008 MacBook is that I forgot to move the filter file to the right directory, when I started listening, I was wondering why it sounded not so clean cut as I was expecting, then I remembered the filter file was not copied and the file directory path was not correct, once I set that up, then it was wonderfull!
My idea is to primarily use UE to linearise the driver performance as much as possible, thus I use only near field measurements as a basis for equalisation, which I find to provide the leanest sound, and also matches my reasoning to do so. There are still some other ideas I have in mind, like trying to do measurements at higher levels to see how much of motor nonlinearity can be captured and how it would effect the listening experience.
I will also take time to get the old ASUS A3500N into the system to see how things perform, but for now, the late 2008 13" MacBook will be the listening reference player where I will also test playing CDs directly. The small system will travel around to different locations to get some feedback from other "really picky people".
Anyway, I am hoping to run at least four channels using the the convolving filter generated by UE. The next test is to run filters of different sample rates because Jriver allows automatic switching of the filters of different sample rates. Currently I use the 96KHs sample rates with the largest buffer size. One advantage of using Jriver for this is that you can concurrently setup different zone names to the same output sound card and switch between the different filters to see if you hear a difference or not. This was a fun experience for me because I can choose between no filter, minimum phase equalisation, phase linearisation only, and the full equalisation.
One occurrence when I just moved to the 2008 MacBook is that I forgot to move the filter file to the right directory, when I started listening, I was wondering why it sounded not so clean cut as I was expecting, then I remembered the filter file was not copied and the file directory path was not correct, once I set that up, then it was wonderfull!
My idea is to primarily use UE to linearise the driver performance as much as possible, thus I use only near field measurements as a basis for equalisation, which I find to provide the leanest sound, and also matches my reasoning to do so. There are still some other ideas I have in mind, like trying to do measurements at higher levels to see how much of motor nonlinearity can be captured and how it would effect the listening experience.
I will also take time to get the old ASUS A3500N into the system to see how things perform, but for now, the late 2008 13" MacBook will be the listening reference player where I will also test playing CDs directly. The small system will travel around to different locations to get some feedback from other "really picky people".
Here are a few things I will be testing first:
1. Equalisation using larger signal measurements. This is to see if we can get the more driver motor nonlinearity into the measurement.
2. Left and Right individual equalisation. For my case, no two drivers are exactly the same, especially in the high frequency end of the driver. My perspective is that driver characteristics down to -40 db or as much as the sample rate limits need to be linearised. The attempt is to make them perform the same as possible.
3. Equalising filters for different sample rates. As Bohdan had mentioned in the manuals, the performance is optimum when you have the music sample rates and the filter sample rates the same.
This process is to help determine the how these changes will effect our listening experience. I have already tested other configurations such as filter based on near field measurement, far field measurement, near field + port measurement. So far I have found near field measurement to be the most clean sound with darkest background perception. The reason is obvious to me, for the far field measurements, you are designing for a single design point, it is highly unlikely that you will have each of your ears at the exact design point, once you are not on the exact design point, the source is actually outputting the inverse of what it was intended to compensate for and arriving at the ear at a different time, thus the sound image will become a bit smeared. When you use near field measurement plus the port, there is also a design point issue where the ears are not on the exact point when listening, plus the fact that your near field measurement may still catch some port output; however, this performs a bit better than the far field measurement where you have the higher frequencies merged with the near field measurement at low frequency and port.
1. Equalisation using larger signal measurements. This is to see if we can get the more driver motor nonlinearity into the measurement.
2. Left and Right individual equalisation. For my case, no two drivers are exactly the same, especially in the high frequency end of the driver. My perspective is that driver characteristics down to -40 db or as much as the sample rate limits need to be linearised. The attempt is to make them perform the same as possible.
3. Equalising filters for different sample rates. As Bohdan had mentioned in the manuals, the performance is optimum when you have the music sample rates and the filter sample rates the same.
This process is to help determine the how these changes will effect our listening experience. I have already tested other configurations such as filter based on near field measurement, far field measurement, near field + port measurement. So far I have found near field measurement to be the most clean sound with darkest background perception. The reason is obvious to me, for the far field measurements, you are designing for a single design point, it is highly unlikely that you will have each of your ears at the exact design point, once you are not on the exact design point, the source is actually outputting the inverse of what it was intended to compensate for and arriving at the ear at a different time, thus the sound image will become a bit smeared. When you use near field measurement plus the port, there is also a design point issue where the ears are not on the exact point when listening, plus the fact that your near field measurement may still catch some port output; however, this performs a bit better than the far field measurement where you have the higher frequencies merged with the near field measurement at low frequency and port.
Last edited:
I recall the first time I used UE V9, I heard something in the music that I could not forget. Later tests never gave me as much an experience. After playing around the options, I discovered that using frequency response of the driver without a mic calibration file provided the unforgettable experience I remembered. I think this is possible that phase shift starts from the recording mic, not using a mic calibration file actually allows the phase linearization process compensate for the recording mic somewhat even though it is not precise.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.