[reter]

Don't obsess over silly things that might not be important or even a real issue at all.
Unless you place the mics exactly the same way into perfectly symmetrical ears(we don't have that), you won't and really shouldn't get identical measurements.
Then there is the headphone and speakers to consider. It's very common to have a few dB here and there on a headphone, as for speakers, the room and placement can have so much impact(and of course the speakers won't have perfectly identical response either).


What you can look for is if you have global imbalance, because it's very unlikely that your own head or mic insertion would cause something with an even impact at all frequencies. So you should probably compensate for that. But even then, maybe your recording setup has one volume level per channel and they're not well matched? Or maybe your amp has some imbalance caused by the volume pot. So it's still relevant to listen and check if the center image(any mono stuff) is well centered for you subjectively. If it is, that's a really good sign that some things are done well.



On that small sample of a capture without scale it's hard to make a guess, but It looks very clean and stable so it's possibly the 50Hz from your house outlet bleeding into the ADC somehow. Doesn't look like noise picked up by the mic(too clean).

Thanks for the answer! This kind of stuff was making me crazy!

I noticed that one of the two mics is less sensitive (the one i plug on the right ear) so i raised the volume gain dx in behringer a little bit, am i doing right? i focus to adjust the red lines on the left to be more or less equal to both mics, is the right thing to do?

This is my last measurement

@lowdown
This is my first time doing the room measurement, i did only one single room measurement (center of the head), just to try the stuff...

but i need to boost the gain to +23db in the windows options to get the 6db headroom, this means that i get a lot of background noise due to the bad amplification the motherboard have, you can hear the noise here


@Xam198 i do the standard general plot with only the headphone files inside the myhrir folder

python impulcifer.py --test_signal="data/sweep-6.15s-48000Hz-32bit-2.93Hz-24000Hz.pkl" --dir_path="data/my_hrir" --plot

 

[Xam198]

Hummm strange, if i don't have the FR-FL file, i get an error from impulcifer

 

[reter]

Hummm strange, if i don't have the FR-FL file, i get an error from impulcifer

maybe i misunderstood, why don't you try doing the whole plot? you can keep the graph open while doing the plot, it refreshes automatically on your windows photo viewer when it's done

 

[Xam198]

Ok i think i found the guilty : seems to be the Zoom H5, i've been able to test without it, with the integrated soundcard, no more zigzag. I think i will command an umc202 !

 

[Xam198]

By the way, i've noticed on my first trials that were not totally Bad that if i increase the volume so that the bass begin to make vibrate the headphones, my brain IS no more cheated by the virtualisation ans the sound comes back around my head. How do you deal with that?

 

[sander99]

By the way, i've noticed on my first trials that were not totally Bad that if i increase the volume so that the bass begin to make vibrate the headphones, my brain IS no more cheated by the virtualisation ans the sound comes back around my head. How do you deal with that?

-Try filtering out ultra low subsonic frequencies
-Maybe attenuate "problem" frequencies that are lifted unreasonably strong as part of the overall speaker simulation (if there are such "problem" frequencies)
-If none of the above works maybe use other headphones that can handle stronger bass without distortion
(-If you use a very weak amp for you headphones try another one)

This actually resembles the only downside I experience when using Sennheiser HD58X and HD600 with the Smyth Realiser A16. But not so much as in your case. In my case it isn't a total collapse of the out-of-head experience, just that a part of the sound starts to sound close to the head. And the feeling that I am not using headphones at all disappears. But it only happens when I play something with a lot of bass at very high levels, louder than I normally do.

 

[Iohfcasa]

Excluding too low frequencies from convolution should prevent from bass distortion, right?

 

[sander99]

Excluding too low frequencies from convolution should prevent from bass distortion, right?

Most "heavy lifting" for the headphone drivers is in the bass. So less deep bass, a lot less work for the drivers, less chance for distortion. At least that is what I am thinking.

 

[reter]

i did my test 9 and WOW, i did 1m13cm far from the speaker, i think i got the best result clarity/distance that's pleasurable for me

For the first time i measured all the 14 steps with the umik-1 with single speaker, oh my what a nightmare! also I'm happy the conservative method of room correction, i think is the less intrusive...

My last hrir is slighly to the right, very very slight, compared to my previus is a huge step forward, sadly the reverb is so much that is still there, less but still and honesly i don't want to sacrifice clarity to "fix" the reverb

 

[morgin]

I’m glad you’re getting good results. It’s awesome when you hear all the little details and positions. Did you do the 7.1? What are you testing with movies/games?

 

[reter]

I’m glad you’re getting good results. It’s awesome when you hear all the little details and positions. Did you do the 7.1? What are you testing with movies/games?

I'm testing mostly videogames and music both stereo and 7.1, yesterday i tried Squad which is free to try on steam this weekend and it has a high dynamic range and WOW this is great!

 

[reter]

Is the room correction already considered when i use the fr_combination_method alone when plotting? or do i have to use the specific_limit command too?

 

[Brandon7s]

Oh man, I haven't visited this thread in months and since I've been so happy with my setup, I hadn't checked back to see how things were going! I have a lot of catching up to do. I'm not sure if anyone has mentioned the cause of this yet, but I ran into this exact same problem quite a few months ago.

Definitely something wrong if you get that zigzag curve. Iirc someone had something similar earlier in the thread but don't remember the solution from the top of my head

Whan you talk about zizag, can you precise wich ones ? The very small ones, oscillating around the curves ?

In my particular case the problem that was causing this zigzag measurement was that the playback of the sine sweep itself was distorting due to being too hot of a signal for my system to handle. I resolved the issue by editing the sweep .wav files directly with Audacity and reducing their amplitude by 2dB. I could also get the distortion to go away by reducing the output volume of windows from 100 to about 70, though I'm not sure about that exact number since it's been so long since I've dealt with this issue.

I'd recommend trying out a sinewave tone generator with a sine set to 50hz and then turn the browser and Windows volume control all the way up (turn your speakers down, that can be LOUD). On my own system, doing this makes the distortion obvious; you'll hear harmonics and overtones that definitely aren't a part of the sinewave. Try lower your Windows/OS volume until you don't hear any distortion, just a pure sinewave tone, and then try running an Impulcifer measurement with your playback volume no higher than that new level. This zigzag problem went away 100% when I did this. I ended up lowering the output on the sweep .wav files just so I could keep my OS volume close to 100% without distortion but lowering OS volume does the same thing, from what I can tell.

 

[Brandon7s]

Does everyone use VBCABLE with EQ-Apo and the listen to option in windows audio settings? Is that the best to get 7.1 surround?

Nah, too much latency for that to be acceptable, in my opinion. I actually ended up buying a cheap 7.1 soundcard that can output via SPDIF and I use that whenever I want 7.1 surround. I've not yet tried the Hifi cable method with ASIO support but this is my first time hearing about it and it sounds like the best way to go. I'll give it a shot sometime this week.

Guys how can i move the left raw and right raw closer? i'm trying multiple times moving changing the mics gain and volume and trying to equalize but i don't know why they don't align in the graph on the right... also no matter how much gain i set, the red line in the two graphs on the left are still almost on top of the Target line

Honestly, your own measurements look GREAT, I don't think you need to get the left and right any tighter in tolerance. That has a relatively minor affect on the results you hear because Impulcifer is automatically compensating for that variance so you don't hear it as being "off" balance. I've had plots that measure +/-7 dB on one side and it still sounded perfectly centered.

For the first time i measured all the 14 steps with the umik-1 with single speaker, oh my what a nightmare! also I'm happy the conservative method of room correction, i think is the less intrusive...

I don't even bother using Impulcifer's room correction - not because it isn't good, but because it's so easy to fix any EQ problems after the measurement is made. I load an EQ VST into EQ-APO (I use Pro-Q 3 but any will work), then I make the necessary corrections while listening to music through the BRIR I'm correcting - and by looking at APO's charts - and then when I am happy with the EQ I burn it to a new .wav file by opening the original in Audacity and applying that same EQ correction. Its MUCH easier to do this than it is for me to go through the whole 14-stage measurement mic process and the results are just as good, in my experience.

My last hrir is slighly to the right, very very slight, compared to my previus is a huge step forward, sadly the reverb is so much that is still there, less but still and honesly i don't want to sacrifice clarity to "fix" the reverb

You can easily adjust the balance issue you mention here (slightly to the right) by processing the BRIR with the --channel_balance=-0.5 or whatever number you want. That parameter adjusts the right channel, so if the right is too loud you can use a minus number to bring it back down so that the left sounds equal, therefor centering the stereo image.

Have you tried using reverb management? If not then I HIGHLY recommend it, it works amazingly well and will definitely reduce the amount of reverb you hear.

Here's an example of what it looks like:
python research\reverberation-management\reverberation_management.py --file="data/my_hrir/69th-Ananda-10K.wav" --track_order=hesuvi --reverb=800

The number at the end is the target length of the resulting reverb tail in milliseconds. 800ms is a good starting point for small rooms, in my opinion.

 

[Xam198]

Oh man, I haven't visited this thread in months and since I've been so happy with my setup, I hadn't checked back to see how things were going! I have a lot of catching up to do. I'm not sure if anyone has mentioned the cause of this yet, but I ran into this exact same problem quite a few months ago.





In my particular case the problem that was causing this zigzag measurement was that the playback of the sine sweep itself was distorting due to being too hot of a signal for my system to handle. I resolved the issue by editing the sweep .wav files directly with Audacity and reducing their amplitude by 2dB. I could also get the distortion to go away by reducing the output volume of windows from 100 to about 70, though I'm not sure about that exact number since it's been so long since I've dealt with this issue.

I'd recommend trying out a sinewave tone generator with a sine set to 50hz and then turn the browser and Windows volume control all the way up (turn your speakers down, that can be LOUD). On my own system, doing this makes the distortion obvious; you'll hear harmonics and overtones that definitely aren't a part of the sinewave. Try lower your Windows/OS volume until you don't hear any distortion, just a pure sinewave tone, and then try running an Impulcifer measurement with your playback volume no higher than that new level. This zigzag problem went away 100% when I did this. I ended up lowering the output on the sweep .wav files just so I could keep my OS volume close to 100% without distortion but lowering OS volume does the same thing, from what I can tell.

Thanks for the advices, but i found (i hope) that my zoom H5 was the cause, because with the integrated soundcard and using the mic in, no more zigzag. But for better qualityt i ordered an umc202hd. We will see.