Jump to content
IGNORED

Inconsistent measurements using JRiver convolution


Recommended Posts

I posted this on the Room Eq Wizard forum but thought perhaps some more experienced people who use both JRMC and REW can chime in. Basically, when using filters with large negative gains, I am seeing very inconsistent measurements in REW even though nothing has changed. I am not sure if this is a JRMC convolver issue or an issue with the filter? Any insight would be greatly appreciated. Here is a portion of what posted on the REW thread (the full post with pics can be found here)

 

Let me explain my setup, I use Jriver as the output device. I use Jriver's convolution so that I can test out my filters to see how they actually perform in my room instead of relying on the predicted curve in REW. In my room, the left speaker as a massive +10db peak at 77hz and the right speaker has a similar peak at 62hz. When I create a filter to flatten out these peaks so they match my target curve, I have to use a -10 gain with a BW60 of 8-10. The result is a very nice flat curve for the left and right speaker. However, when I measure with both speakers I get inconsistent results. Sometimes I get the massive nulls precisely at where I had created a filter to deal with the large peaks for the left and right speaker. Other times I get a smooth curve with no nulls but with a large trough between 60-80hz. When I do a measurement with convolution turned off in JRiver the nulls at 62hz and 77hz goes away.

 

So clearly Jriver's convolution engine is doing something strange with the filter's I created. Why would I get a flat curve when only measuring 1 speaker but get inconsistent results with both speakers? Also, when I do get measurements that result in a trough versus 2 nulls, the trough actually dips below the db level of each individual speaker. How is that possible?

12TB NAS >> i7-6700 Server/Control PC >> i3-5015u NAA >> Singxer SU-1 DDC (modded) >> Holo Spring L3 DAC >> Accustic Arts Power 1 int amp >> Sonus Faber Guaneri Evolution speakers + REL T/5i sub (x2)

 

Other components:

UpTone Audio LPS1.2/IsoRegen, Fiber Switch and FMC, Windows Server 2016 OS, Audiophile Optimizer 3.0, Fidelizer Pro 6, HQ Player, Roonserver, PS Audio P3 AC regenerator, HDPlex 400W ATX & 200W Linear PSU, Light Harmonic Lightspeed Split USB cable, Synergistic Research Tungsten AC power cords, Tara Labs The One speaker cables, Tara Labs The Two Extended with HFX Station IC, Oyaide R1 outlets, Stillpoints Ultra Mini footers, Hi-Fi Tuning fuses, Vicoustic/RealTraps/GIK room treatments

Link to comment

You may be getting phase cancellation at the microphone. If you take each speaker curve and sum them in REW, do you get the desired results? How does it sound? I have experienced the same issue when doing measurements of both speakers together.

Main System: [Synology DS216, Rpi-4b LMS (pCP)], Holo Audio Red, Ayre QX-5 Twenty, Ayre KX-5 Twenty, Ayre VX-5 Twenty, Revel Ultima Studio2, Iconoclast speaker cables & interconnects, RealTraps acoustic treatments

Living Room: Sonore ultraRendu, Ayre QB-9DSD, Simaudio MOON 340iX, B&W 802 Diamond

Link to comment

You could also try checking the results of your curve by using pink noise. I believe that REW has a RTA feature w/ pink noise, this should give you a ballpark idea of what things look like, i.e., is there a huge null or not.

Main System: [Synology DS216, Rpi-4b LMS (pCP)], Holo Audio Red, Ayre QX-5 Twenty, Ayre KX-5 Twenty, Ayre VX-5 Twenty, Revel Ultima Studio2, Iconoclast speaker cables & interconnects, RealTraps acoustic treatments

Living Room: Sonore ultraRendu, Ayre QB-9DSD, Simaudio MOON 340iX, B&W 802 Diamond

Link to comment
You may be getting phase cancellation at the microphone. If you take each speaker curve and sum them in REW, do you get the desired results? How does it sound? I have experienced the same issue when doing measurements of both speakers together.

WOW! I never used the sum feature in REW. I am amazed how closely the summed curve resembles one of the actual measured curve! Ok, the summed curve has the deep dips at 62 and 77hz. Is this just an abnormality due to the large, wide negative gain I applied at those points? I dont see these same big dips when I measure each speaker by itself with the filter applied.

You could also try checking the results of your curve by using pink noise. I believe that REW has a RTA feature w/ pink noise, this should give you a ballpark idea of what things look like, i.e., is there a huge null or not.

 

The difficult thing is that I am trying to verify the response with the filters. When I use the REW signal generator and SPL, I get inconsistent results when I apply my filters through JRMC convolver. Sometimes the I can verify the null with the SPL and my ears, other times there is no null.

12TB NAS >> i7-6700 Server/Control PC >> i3-5015u NAA >> Singxer SU-1 DDC (modded) >> Holo Spring L3 DAC >> Accustic Arts Power 1 int amp >> Sonus Faber Guaneri Evolution speakers + REL T/5i sub (x2)

 

Other components:

UpTone Audio LPS1.2/IsoRegen, Fiber Switch and FMC, Windows Server 2016 OS, Audiophile Optimizer 3.0, Fidelizer Pro 6, HQ Player, Roonserver, PS Audio P3 AC regenerator, HDPlex 400W ATX & 200W Linear PSU, Light Harmonic Lightspeed Split USB cable, Synergistic Research Tungsten AC power cords, Tara Labs The One speaker cables, Tara Labs The Two Extended with HFX Station IC, Oyaide R1 outlets, Stillpoints Ultra Mini footers, Hi-Fi Tuning fuses, Vicoustic/RealTraps/GIK room treatments

Link to comment

Can't explain the inconsistency. Maybe I am not following 100%. Do you have a CD with pink noise or warble tones (Stereophile test CD) that you can rip into JRMC. Playing back the tones from JRMC should let you see the effect of the filters. Sorry, don't know what to suggest beyond this.

Main System: [Synology DS216, Rpi-4b LMS (pCP)], Holo Audio Red, Ayre QX-5 Twenty, Ayre KX-5 Twenty, Ayre VX-5 Twenty, Revel Ultima Studio2, Iconoclast speaker cables & interconnects, RealTraps acoustic treatments

Living Room: Sonore ultraRendu, Ayre QB-9DSD, Simaudio MOON 340iX, B&W 802 Diamond

Link to comment
Can't explain the inconsistency. Maybe I am not following 100%. Do you have a CD with pink noise or warble tones (Stereophile test CD) that you can rip into JRMC. Playing back the tones from JRMC should let you see the effect of the filters. Sorry, don't know what to suggest beyond this.

My apologies for not being clear. I really appreciate you trying to help.

 

I can generate sine wave test tones with Room EQ Wizard which is set up to send the output through JRMC. My laptop is then plugged into my receiver. By sending the signal through JRMC and enabling convolution, I can then use REW to measure the effect of the filters.

 

Someone on the JRMC forums suggested that I copy the filter value into the parametric equalizer settings and disable convolution to see if I get similar or different results. If the results are the same then its unlikely that JRMC convolution engine is doing something wonky. Will report back shortly once I get the equalizer settings entered into JRMC and have some measurements.

12TB NAS >> i7-6700 Server/Control PC >> i3-5015u NAA >> Singxer SU-1 DDC (modded) >> Holo Spring L3 DAC >> Accustic Arts Power 1 int amp >> Sonus Faber Guaneri Evolution speakers + REL T/5i sub (x2)

 

Other components:

UpTone Audio LPS1.2/IsoRegen, Fiber Switch and FMC, Windows Server 2016 OS, Audiophile Optimizer 3.0, Fidelizer Pro 6, HQ Player, Roonserver, PS Audio P3 AC regenerator, HDPlex 400W ATX & 200W Linear PSU, Light Harmonic Lightspeed Split USB cable, Synergistic Research Tungsten AC power cords, Tara Labs The One speaker cables, Tara Labs The Two Extended with HFX Station IC, Oyaide R1 outlets, Stillpoints Ultra Mini footers, Hi-Fi Tuning fuses, Vicoustic/RealTraps/GIK room treatments

Link to comment

More testing...I manually entered the REW filter settings into JRMC parametric eq and reran the measurements. The combined graphs show the double dip at 62 and 77hz (precisely where I have my negative gain filter on the R and L speaker respectively). The measurements for each speaker is also very similar to when I use JRMC convolution engine. Therefore, I have to believe the inconsistent results of the combined measurement has to do with the JRMC convolution engine.

Regarding the phase...I noticed that when I have all filters off and just measure individual frequencies in the range of 60-80hz, the measured combined output is about 4-5db less than the individual levels. For example, at 70hz the left and right speakers measure 77db but the combined output is 73db. So does this mean I have a slight phase issue in this range? If so, is there a way I can fix it?

12TB NAS >> i7-6700 Server/Control PC >> i3-5015u NAA >> Singxer SU-1 DDC (modded) >> Holo Spring L3 DAC >> Accustic Arts Power 1 int amp >> Sonus Faber Guaneri Evolution speakers + REL T/5i sub (x2)

 

Other components:

UpTone Audio LPS1.2/IsoRegen, Fiber Switch and FMC, Windows Server 2016 OS, Audiophile Optimizer 3.0, Fidelizer Pro 6, HQ Player, Roonserver, PS Audio P3 AC regenerator, HDPlex 400W ATX & 200W Linear PSU, Light Harmonic Lightspeed Split USB cable, Synergistic Research Tungsten AC power cords, Tara Labs The One speaker cables, Tara Labs The Two Extended with HFX Station IC, Oyaide R1 outlets, Stillpoints Ultra Mini footers, Hi-Fi Tuning fuses, Vicoustic/RealTraps/GIK room treatments

Link to comment
More testing...I manually entered the REW filter settings into JRMC parametric eq and reran the measurements. The combined graphs show the double dip at 62 and 77hz (precisely where I have my negative gain filter on the R and L speaker respectively). The measurements for each speaker is also very similar to when I use JRMC convolution engine. Therefore, I have to believe the inconsistent results of the combined measurement has to do with the JRMC convolution engine.

Regarding the phase...I noticed that when I have all filters off and just measure individual frequencies in the range of 60-80hz, the measured combined output is about 4-5db less than the individual levels. For example, at 70hz the left and right speakers measure 77db but the combined output is 73db. So does this mean I have a slight phase issue in this range? If so, is there a way I can fix it?

 

It has been a long time since I've used REW (it isn't great on a Mac). What you are seeing may be a phase issue but it may also be an artifact of where the mic is positioned, i.e., you're two ears may not hear it that way. Two suggestions: I assume that you are measuring each speaker separately (77dB values) and summing them in REW (73dB value). My guess is that if you measure both speakers at the same time, you will also see the reduced output. Is there a way of stripping the phase information from the measurements before summing (this is something that I used to do in FuzzMeasure)? If you can do this, does the sum make more sense? Second, check your response with pink noise or warble tones, these tones tend to not have phase issues. Please note, I am not an expert with this stuff, I can only share what I learned when measuring my old listening room. Phase cancellation was a huge issue at my "problem" frequency. The only way I could get accurate measurements was to measure each speaker separately, strip out the phase information, and then sum (summing should increase level by 3 dB, averaging keeps levels the same). I could confirm the measurement by using pink noise or warble tones. So a very big YMMV. :)

Main System: [Synology DS216, Rpi-4b LMS (pCP)], Holo Audio Red, Ayre QX-5 Twenty, Ayre KX-5 Twenty, Ayre VX-5 Twenty, Revel Ultima Studio2, Iconoclast speaker cables & interconnects, RealTraps acoustic treatments

Living Room: Sonore ultraRendu, Ayre QB-9DSD, Simaudio MOON 340iX, B&W 802 Diamond

Link to comment
It has been a long time since I've used REW (it isn't great on a Mac). What you are seeing may be a phase issue but it may also be an artifact of where the mic is positioned, i.e., you're two ears may not hear it that way. Two suggestions: I assume that you are measuring each speaker separately (77dB values) and summing them in REW (73dB value). My guess is that if you measure both speakers at the same time, you will also see the reduced output. Is there a way of stripping the phase information from the measurements before summing (this is something that I used to do in FuzzMeasure)? If you can do this, does the sum make more sense? Second, check your response with pink noise or warble tones, these tones tend to not have phase issues. Please note, I am not an expert with this stuff, I can only share what I learned when measuring my old listening room. Phase cancellation was a huge issue at my "problem" frequency. The only way I could get accurate measurements was to measure each speaker separately, strip out the phase information, and then sum (summing should increase level by 3 dB, averaging keeps levels the same). I could confirm the measurement by using pink noise or warble tones. So a very big YMMV. :)

Yes, I measure each speaker separately, then I measure them together as well. I am not using REW to sum them.

 

I will need to look into the phase information available in REW. I am total noob when it comes to this stuff so I am learning as go along.

 

Thank you so much!

12TB NAS >> i7-6700 Server/Control PC >> i3-5015u NAA >> Singxer SU-1 DDC (modded) >> Holo Spring L3 DAC >> Accustic Arts Power 1 int amp >> Sonus Faber Guaneri Evolution speakers + REL T/5i sub (x2)

 

Other components:

UpTone Audio LPS1.2/IsoRegen, Fiber Switch and FMC, Windows Server 2016 OS, Audiophile Optimizer 3.0, Fidelizer Pro 6, HQ Player, Roonserver, PS Audio P3 AC regenerator, HDPlex 400W ATX & 200W Linear PSU, Light Harmonic Lightspeed Split USB cable, Synergistic Research Tungsten AC power cords, Tara Labs The One speaker cables, Tara Labs The Two Extended with HFX Station IC, Oyaide R1 outlets, Stillpoints Ultra Mini footers, Hi-Fi Tuning fuses, Vicoustic/RealTraps/GIK room treatments

Link to comment
I posted this on the Room Eq Wizard forum but thought perhaps some more experienced people who use both JRMC and REW can chime in. Basically, when using filters with large negative gains, I am seeing very inconsistent measurements in REW even though nothing has changed. I am not sure if this is a JRMC convolver issue or an issue with the filter? Any insight would be greatly appreciated. Here is a portion of what posted on the REW thread (the full post with pics can be found here)

Let me explain my setup, I use Jriver as the output device. I use Jriver's convolution so that I can test out my filters to see how they actually perform in my room instead of relying on the predicted curve in REW. In my room, the left speaker as a massive +10db peak at 77hz and the right speaker has a similar peak at 62hz. When I create a filter to flatten out these peaks so they match my target curve, I have to use a -10 gain with a BW60 of 8-10. The result is a very nice flat curve for the left and right speaker. However, when I measure with both speakers I get inconsistent results. Sometimes I get the massive nulls precisely at where I had created a filter to deal with the large peaks for the left and right speaker. Other times I get a smooth curve with no nulls but with a large trough between 60-80hz. When I do a measurement with convolution turned off in JRiver the nulls at 62hz and 77hz goes away.

 

So clearly Jriver's convolution engine is doing something strange with the filter's I created. Why would I get a flat curve when only measuring 1 speaker but get inconsistent results with both speakers? Also, when I do get measurements that result in a trough versus 2 nulls, the trough actually dips below the db level of each individual speaker. How is that possible?

 

I am afraid your room is too small ie shorter than the wave length of 77Hz (3.88 metres). Room equalization won't help you. A null occurs when sound waves meet their own reflections 180 degrees out of phase. It's a physical phenomenon for which there is no electronic cure. Bass traps down to these frequencies have to be very large (not convenient in a small room). A Helmholtz resonator is the only solution, and even that wouldn't fix the problem completely(you can hang them from the ceiling to save space). Also try changing your sitting position to the geometrical centre of the room.

 

However if you still want to play around with room EQ the link below is an excellent source. There's some detailed information on how to use JRiver with REW and also some close-micing measurement techniques by using HOLM(also free) which has frequency dependent windowing feature which REW lacks:

 

https://yabb.jriver.com/interact/index.php?topic=87538.0

Link to comment

tboooe, so I downloaded your baseline MDAT files to take a look in REW. I have also read the comments people made and I'm in general agreement with everyone.

 

First, I do have a question. How did you integrate your left subwoofer with your left speaker and the right subwoofer with the right speaker?

 

Second, the comment is that it's just going to be impossible to EQ both channels individually to have a flat frequency response because once you combine them, because of phase issues, the speakers and subwoofers are going to interfere with each other so your combined speaker channel responses are going to generate some dips and peaks. That's why your untreated combined L+R speaker frequency responses actually look flatter than the individual speaker responses. I usually just live with this. Fortunately, bass is less directional and usually both channels have similar bass volumes (but not necessarily) so most of the time, the difference in volume is not completely devastating (even if not ideal).

 

Third, the reason why multiple measurements get different results, particularly when you're using both speakers, as people have said is because of phase cancellation. And phase cancellation effects are most prominent around room resonances. Slight changes to the position of the microphone can have a big impact. Ideally, for the untreated frequency response measurements, you really want to measure at multiple spots to get a sense of what you really need to do to optimize the sound. I usually measure at my left ear, right ear, in the centre and then look at the individual curves and the combined averaged curve. And I would do that for the left channel, the right channel and the combined left + right together before deciding whether I can use parametric EQ to improve the sound further.

 

Finally, assuming your original baseline measurements are representative of the general frequency response of your seating position, I would say that your left channel 77Hz peak is generally cancelled or balanced out by the right channel dip/phase cancellation and your right channel 62Hz peak is generally cancelled or balanced out by the left channel dip/phase cancellation. As a result, if everything is setup perfectly in your system (assuming proper subwoofer integration), I would not use any parametric EQ to correct for the 62Hz & 77Hz issues. I think it's just something to live with. What I would do is to remove the slight peak with a parametric EQ to L+R channels at 58.1Hz, -3dB, Q=12. This is more of an art of EQ and then re-listen to all your music and see if you enjoy the experience more. If it's still problematic, you might want to increase the dip to say -4dB or change the Q slightly.

Link to comment

Just one more additional comment. I know that from the other forum, you integrated your subwoofer with a crossover at 40Hz and the speakers running at full range. And I agree the frequency and phase response looks great from that standpoint. And you said your target curve is to get 20-100Hz to be as flat as possible. Except from experience, most people don't listen at concert level volumes. As a result, a slight increase in volume in the 20-40Hz range usually sounds psychoacoustically better. I have to admit, looking at your frequency response curves, even taking into account your preferences, I would probably crank up the subwoofer volume by 2-3 dB and then lower the crossover frequency slightly to compensate for the increase in volume and then double checking to make sure the phase and frequency response is still reasonably smooth. It should give a fuller more 3-dimensional soundstage.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...