Recording Impulse Responses for Speaker Virtualization

[morgin]

I miss the updates from you guys. Anyone tried anything new to get better results?

 

[Brandon7s]

I miss the updates from you guys. Anyone tried anything new to get better results?

I haven't made any new measurements in a few weeks, not since the last measurement I posted about here. I still intend to make one with my pair of LSR 308s to see how they fare against my Kali LP-6s that I recently bought, but I've been so happy with my last set of measurements that I've not felt any need to make some more.

That said, this morning I just glued my master series Sound Professional binaural mics (that I removed the silicone housing from) to some foam earplugs and I'm going to make take another measurement with these soon. My last few measurements had been without foam earplugs, I was simply putting them a little ways into my ears without any covering or anything, but I feel like that's made for an unnaturally wide stereo image on the BRIR. As a result I have been using the avg channel_balance and that's worked very well for me. It does change the EQ of the BRIR a bit, making it darker and more muffled, but I've been using some post-BRIR EQ on the highs to fix that. I think that putting them directly in the ear makes the mic measurements inconsistent since their orientation isn't fixed in place, so putting them on foam earplugs again will likely fix that.

I've been hesitant to glue them directly to foam since this isn't a reversible modification. I thought about putting some tape on them and then gluing the tape, but there's so little surface area without the housing that it just doesn't stick enough to work well. I think this will be fine though, it won't make the mics any more mismatched than they were prior to removing the casing, I'm pretty confident of that.

I'll take some more measurements this weekend, including with the JBL LSR 308s, and will report back on the results.

 

[lowdown]

I've been so happy with my last set of measurements that I've not felt any need to make some more.

I've been hesitant to glue them directly to foam since this isn't a reversible modification.

I'd like to have a $ for every time I've thought while listening "I can't imagine how this could sound better." I do think about taking more measurements sometimes, but then I listen again and my motivation evaporates.

A suggestion on the non-reversible glue issue, I used silicone caulk for that very reason. Once it sets it's secure, but also flexible enough to be removed if desired without damaging the mics.

 

[morgin]

I've been hesitant to glue them directly to foam since this isn't a reversible modification

My best measurements were with foam. I’ve tried them naked in my ear and with silicone housing. But the best have been with foam.

I cut out a little ditch in the middle where the mic would be glued but made sure the mic isn’t flush with the foam. And used a hot glue gun because I find that easily peels off if I need to remove it.

Really wanna try getting a sound treated room to rent to see if that makes a big impact. But need to find reliable consistent recordings first

 

[Brandon7s]

I'd like to have a $ for every time I've thought while listening "I can't imagine how this could sound better." I do think about taking more measurements sometimes, but then I listen again and my motivation evaporates.

Very true! I actually just straight-up prefer using my Ananda's with my latest BRIR over using my speakers now. And not just because I don't have to worry about annoying the neighbors by being too loud. The acoustic problems with my room stick out like a sore thumb to me since I'm so used to listening with a BRIR that has all of those issues corrected with EQ. And my speakers can't produce sub-bass like my headphones can so they speakers seem thin and weak by comparison.

A suggestion on the non-reversible glue issue, I used silicone caulk for that very reason. Once it sets it's secure, but also flexible enough to be removed if desired without damaging the mics.

That's brilliant! I'm going to pick some up this weekend. I never would have thought of that in a million years.

I cut out a little ditch in the middle where the mic would be glued but made sure the mic isn’t flush with the foam. And used a hot glue gun because I find that easily peels off if I need to remove it.

Really wanna try getting a sound treated room to rent to see if that makes a big impact. But need to find reliable consistent recordings first

The hot glue also sounds like a good option. I'd love to try making a measurement in an acoustically treated room as well, though I find myself wondering how much of a difference that would really make vs. correcting BRIRs with EQ like I've been doing. It probably would be an improvement but just how much of an improvement would be very interesting to discover.

 

[lowdown]

Very true! I actually just straight-up prefer using my Ananda's with my latest BRIR over using my speakers now. And not just because I don't have to worry about annoying the neighbors by being too loud. The acoustic problems with my room stick out like a sore thumb to me since I'm so used to listening with a BRIR that has all of those issues corrected with EQ. And my speakers can't produce sub-bass like my headphones can so they speakers seem thin and weak by comparison.

I actually recently sold the fancy speakers I'd had for years on eBay. My headphones with the Impulcifer enhancement are so much better than listening to the speakers that they became just expensive decorations. All the issues with my room and speaker anomalies I spent so much time trying to tweak have been totally eliminated. And with the speakers I would always be worried about whether my neighbors were home, the time of day, and whether the volume was perhaps a bit too loud. The complete freedom to listen at any time at the ideal volume, and to have as close to perfect sound as I can imagine after seeking it for so long, is more than amazing.

 

[Iohfcasa]

I have used pritt sticky tacks, because they were laying around here, they stick well to the capsules and are even moldable like plasticine.

 

[musicreo]

I have tested the option to equalize for a different headphone. I tried to equalize my old HD555 measurement to a HD600 using the instructions in the readme. The result is that my own equalizer setting I made sounds much better for me.

 

[Brandon7s]

Will you also give your second pair of jbl 305 a try with the more even mic position?

Sorry for being so late on getting a new measurement with my LSR 308s, I just finally did that and the results turned out fantastic. I think this is my favorite BRIR yet, though that's probably just as much a result of a change to the mic position as it is to the speakers. There's absolutely nothing wrong with these 308s, in any case! They do have a little more definition in the low end compared to my Kali LP-6s, but they also do seem to exaggerate a particularly bad standing wave in my room at about 105hz. That issue is lessened with the LP-6s but not fixed completely by any stretch; the difference is not significant.

So yeah, the LSR 308 first generation speakers get a big thumbs up from me, and I bet that a lot of the channel balance problems I had while using my 305s were from a fault in one of speakers themselves. I don't think there's any inherent issue in the design of the speakers themselves that was making good measurements difficult. This latest measurement has the most accurate imaging and channel balance that I've gotten so far.

One problem that I've been having is that the mic capsules get turned and twisted from their neutral position on the foam earplugs due to the cable not being flexible enough and forcing the capsule to tilt. Here's what I did to fix that:

I cut a slit to allow the cable to flow down more freely from the capsule so it wouldn't tilt it and then I just glued the cable to the foam in a small loop leading back to the top side. It works great! The capsules lay completely flat in relation to the foam, keeping them both much more centered and evenly placed related to each ear's structure while being worn. I've not taken a measurement with my Kali monitors with this yet but the 308 measurement I made with this has absolutely no problems with the stereo image being off-centered or unnaturally wide, which was something that was so inconsistent before. Once I make a measurement with these on my Kali LP-6s I'll be able to more accurately compare how I like them vs. my LSR 308s. I'll get that done sometime this week.

Also, I noticed something a bit odd today and I wonder if anyone else has experienced it. The last few measurements I've taken have been with generic room measurements to correct the EQ through Impulcifer, but all of those have a weird low end ringing, though I'm not sure if that's even the right word... it's like the very low end is reverberating longer than the rest of the frequencies but mostly only in one ear. It almost sounds like a slapback delay limited to the bass and sub-bass region.

When I process the same measurements using --no_room_correction it sounds perfectly fine and the low-end ringing/delay is nowhere to be found. At first I thought that means there's some low-end issues in my room that are being brought to the forefront and becoming noticeable simply because those low frequencies were too low in amplitude to be heard unless they were boosted via EQ; however, I just did some manual EQing to obtain a close match for the Impulcifer EQ results and the problem is completely absent. I've not yet tried processing the BRIR while using the Average or Conservative processing modes for the room correction but I'll try both of those and see if the problem goes away when using one versus the other. I've been taking 3 generic room measurements using a calibrated microphone, one at center of where the head would be, one at slightly right, and one slightly left.

This isn't a big issue though, I can generate a measurement of my manual EQ corrections using REW and then simply toss the resulting .csv file into the my_hrir folder to bake the EQ directly into the BRIR itself, which will save additional processing further on down the line. Though I've not actually tried that yet... hopefully it doesn't bring back the ringing/delay problem. I'll report back once I've gotten the time to try it.

 

[musicreo]

@Brandon7s​

Have you tried if the decay option helps with your problem?

 

[Brandon7s]

@Brandon7s​

Have you tried if the decay option helps with your problem?

I actually haven't tried the decay option for this yet but I HAVE tried using reverbation management and that helps significantly. The ringing/delay goes away at about 600ms and 700ms for most of the measurements I've run into that have this issue when using room correction through Impulcifer.

 

[conql]

I have been digging into this idea for months without knowing Impulcifer and was about to write a program on my own to implement it. But then I found the Impulcifer, which has such a good completion and elegant algorithms! It is such an amazing job! Can't say how much I love it!

Has anyone made the silicone mold microphones proposed by Fabian Brinkmann? According to his paper, this yields the least inter-measurement variance.

I made a pair of them following his idea but with different types of silicone. This is what it looks like. Though I don't have the equipment to measure its variance, it does seem quite stable during multiple measurements.

The mics are bought from a Chinese company called "Yinglear", costing less than one dollar each. They claim that it has 72+dB SNR and flat frequency response. I don't know whether it's exactly true, but it does have an SNR greater than 65dB from my experience.

This thread is too long and I haven't read through it yet. I wonder has anyone experienced the loss of precision for high frequency (10k+) measurement? Plus, I do feel very bright in 10k+ when comparing the simulation to real loudspeakers.

https://github.com/jaakkopasanen/Impulcifer/issues/61

Sooooo glad to see this thread.

 

[morgin]

Hello and welcome, it’s always nice to see more people discovering this treasure and more input should help everyone get good results.

Has anyone made the silicone mold microphones proposed by Fabian Brinkmann? According to his paper, this yields the least inter-measurement variance.

This is interesting and something new to try out do you have a link to this paper?

Edit found the paper if anyone else wants to check it out

https://www2.ak.tu-berlin.de/~akgroup/ak_pub/abschlussarbeiten/2011/Brinkmann_MagA.pdf


Also wanted to ask if it was better in anyway to have multiple measurements and then have impulcifer average them out to get the best results from all recordings?

 

[musicreo]

The mics are bought from a Chinese company called "Yinglear", costing less than one dollar each. They claim that it has 72+dB SNR and flat frequency response. I don't know whether it's exactly true, but it does have an SNR greater than 65dB from my experience.

Less than one dollar is great. The Primo Em258 is ten times that price.

This thread is too long and I haven't read through it yet. I wonder has anyone experienced the loss of precision for high frequency (10k+) measurement? Plus, I do feel very bright in 10k+ when comparing the simulation to real loudspeakers.

The accuracy for headphones measurements is a problem at high frequencies. The best way would be to do the measurements at the eardrum with a probe mic. But this is not possible with our kind of microphone capsules.
For me I would say that blocked ear channal measurements sound more bright while measurements with only partial blocked ear channal sound better.

 

[conql]

Also wanted to ask if it was better in anyway to have multiple measurements and then have impulcifer average them out to get the best results from all recordings?

Theoretically, doubling the number of averages increases the SNR of the measurement result by 3 dB, and lengthening the sweep signal can do that too. In my experience, lengthening the sweep signal does improve high-frequency accuracy for measuring headphones with Impulcifer **, but I haven't tested that in measuring loudspeakers.

For me I would say that blocked ear channal measurements sound more bright while measurements with only partial blocked ear channal sound better.

Thanks, I will give that a try.

Edit:
**: This is not true. A longer sweep does not significantly increase high-frequency accuracy. The default 6.15s sweep is sufficient already.