[johnn29]

Watched a few shows/movies with 300ms on a couple of my HRIR's - no ill effects.

Think it's ready for the main branch!

 

[lowdown]

A couple of questions. Sorry if they are too basic, or I'm so confused even the questions don't make sense.

What's the difference between the hesuvi.wav and hrir.wav files? I've been using hesuvi.wav. What is hrir.wav for?

The doc says "Impulcifer will compensate for the headphone frequency response using headphone sine sweep recording". Is there a default eq curve that you compensate the headphone/ear measurement to match? I realize there are ways to bake in other eq adjustments as documented, but I'm wondering what the target is without the extra eq.csv file.

Does the "results" plot show the L/R FR curves that I'm hearing at my ear canal, so it should match my ideal FR curve, or is that the FR curve that is sent to the headphone that then gets modified by my specific headphone response curve and ear pinna? I'm trying to understand from the results plot how far off I am from my ideal curve. Maybe the "pre" and "post" plots are closer to what I'm after but I can't tell from the documentation.

Thanks.

 

[musicreo]

A couple of questions. Sorry if they are too basic, or I'm so confused even the questions don't make sense.

What's the difference between the hesuvi.wav and hrir.wav files? I've been using hesuvi.wav. What is hrir.wav for?

Hrir.wav has a different channel order compared to Hesuvi.wav. The reason is the very strange channel configuration in hesuvi.

The doc says "Impulcifer will compensate for the headphone frequency response using headphone sine sweep recording". Is there a default eq curve that you compensate the headphone/ear measurement to match? I realize there are ways to bake in other eq adjustments as documented, but I'm wondering what the target is without the extra eq.csv file.

In the moment the plot shows a flat target. So I guess that's it.

 

[jaakkopasanen]

Left curve in results plot is the average frequency response of all left ear impulse responses. The results plot contains all the equalizations, corrections and what not. It's the total EQ curve that gets sent to the headphones.

I'm not sure how the results curve relates to an the personal ideal FR target without speaker virtualization. Maybe it would be good with HRIR measurements that only have FL and FR channels. Also could be that ideal target for virtual speakers is a bit different than for headphones without virtualization.

 

[lowdown]

Hrir.wav has a different channel order compared to Hesuvi.wav. The reason is the very strange channel configuration in hesuvi.

Ok, got it. Thanks.

Left curve in results plot is the average frequency response of all left ear impulse responses. The results plot contains all the equalizations, corrections and what not. It's the total EQ curve that gets sent to the headphones.

I'm not sure how the results curve relates to an the personal ideal FR target without speaker virtualization. Maybe it would be good with HRIR measurements that only have FL and FR channels. Also could be that ideal target for virtual speakers is a bit different than for headphones without virtualization.

That helps, but I'm still unclear, which I know is my lack of knowledge. Here's my basic problem, and maybe I should have presented it this way to start. I'm only dealing with stereo, FL/FR for now. If I just listen to music with my speakers and on some tracks I hear something out of tonal balance, is it the recording, or the FR of my speakers/amp/room? In that case I can use my calibrated UMIK-1 and REW and see the plot of the FR of my equipment at my listening point. I can then make adjustments to get the curve closer to what I want without guessing where a specific peak or drop may be. Is that too much thump I hear in the bass on that track in the recording or my system, and if it's my system which freq needs adjusting? I look at the results plot for my hesuvi.wav and I see large 10dB swings up and down along the frequency range. I can't figure out how to intepret that, whether it's normal and good, or really abnormal. Using Impulcifer with my headphones is as if all of my recordings have been remastered. As I've said before it's amazingly good, but the freq balance has changed, and I can't tell if it's that much more true to the recordings or if it's introduced an imbalance that I could fix with eq. I know some people are experts at this but for me using individual music tracks to take stabs at where and how much to tweak the eq on specific frequencies is a pretty hit and miss approach. Is there a procedure or a way of interpreting plots that would help me narrow this down?

 

[johnn29]

lowdown - are you using the virtual room correction? That'll handle everything automagically and EQ to what's been shown to be a listening panel's preference - the Harman Room curve. If you are using that - that is why it sounds different.

A good frequency response in room would be +/- 5db - the virtual room correction can you you close to that.

I know humans find it hard to pick up on frequency nulls more than booms.

For what it's worth in my entire "career" of being an audio-nerd I've never put much stock into a flat frequency response. Your brain copes with much of it automatically. Now I have Impulcifer I can generate a flatter frequency response vs the natural room and A vs B. I couldn't really notice much difference outside of the bass boost. But it satisfies the audio-OCD in me to know that I have a much flatter frequency response than in my real room with the virtual room.

 

[lowdown]

lowdown - are you using the virtual room correction? That'll handle everything automagically and EQ to what's been shown to be a listening panel's preference - the Harman Room curve. If you are using that - that is why it sounds different.

A good frequency response in room would be +/- 5db - the virtual room correction can you you close to that.

I know humans find it hard to pick up on frequency nulls more than booms.

For what it's worth in my entire "career" of being an audio-nerd I've never put much stock into a flat frequency response. Your brain copes with much of it automatically. Now I have Impulcifer I can generate a flatter frequency response vs the natural room and A vs B. I couldn't really notice much difference outside of the bass boost. But it satisfies the audio-OCD in me to know that I have a much flatter frequency response than in my real room with the virtual room.

Yes, I'm using room correction. I've tried both with the Harmon curve, and with versions of it that I've tweaked. The first time I used it stock the bass was boosted too much, so I've created a couple of edited versions. I understand that flat is not most people's preference. I'm not stuck trying to achieve flat, but what's challenging for me is determining where and how much to use eq, or some other option, to get closer to my ideal. I'm getting a lot of variation in sound quality from my final resulting hesuvi.wav files after several different measuring sessions. And I'm having trouble isolating which step or combination is causing so much variation. Then when I get a result that's very good but not quite ideal how do I make the right adjustment? Part of it is inconsistency in my measurements, both with headphones on and room measurements, and part is having so many options to tweak in the last computation command line. Then lastly I end up trying to make eq adjustments by ear using various music tracks with no objective measurement feedback.

I totally agree that Impulcifer is better than my speakers. I've basically stopped listening to music through my speakers because the headphone experience is so much more revealing and captivating, and the overall sound quality is so good. But there are still tonal and spacial anomalies that I'd like to improve and am working my way through how to do that.

 

[johnn29]

Ah I see what you mean. You can generate different virtual room targets that Jaako tried (https://github.com/jaakkopasanen/Impulcifer/tree/master/research/virtual-room-target). See if you like any of those better.

In my experience with trying to figure out what HRIR sounds good I was even thrown off by being in different rooms. I posted about it when I first generated a good HRIR that I was very happy with when listening in my home theater. I then took it into a cafe and found it sounded terrible and that Dolby Headphone sounded far better. It's because my brain wasn't expecting the theater sound in a loud and novel environment.

I'd say the easiest thing to do is shoot for the harman target and then use simple bass/mid/treble tone controls on some of your favorite tracks to see what you like. When Harman's studied room correction systems they found that listeners didn't actually prefer them and simple tone controls were sufficient anyway.

In some ways having a HRIR means we can do A vs B so much quicker and easier than on a real loud speaker setup. But in other ways because you can switch so quickly between HRIR's you don't give your brain enough time to compensate. You then end up fiddling with many settings. For my own listening i've created 3 go-to HRIR's. One is nearfield that I use when close to a screen - like a laptop or desktop monitor, the other is midfield where I'm sitting 1-2m away from a TV and the final is far-field whic his for projection use. The only one I can consistently use in all situations is nearfield - if I use the far field one in a different environment it just sounds wrong.

 

[lowdown]

Ah I see what you mean. You can generate different virtual room targets that Jaako tried (https://github.com/jaakkopasanen/Impulcifer/tree/master/research/virtual-room-target). See if you like any of those better.

In my experience with trying to figure out what HRIR sounds good I was even thrown off by being in different rooms. I posted about it when I first generated a good HRIR that I was very happy with when listening in my home theater. I then took it into a cafe and found it sounded terrible and that Dolby Headphone sounded far better. It's because my brain wasn't expecting the theater sound in a loud and novel environment.

I'd say the easiest thing to do is shoot for the harman target and then use simple bass/mid/treble tone controls on some of your favorite tracks to see what you like. When Harman's studied room correction systems they found that listeners didn't actually prefer them and simple tone controls were sufficient anyway.

Ok, I've looked at that but don't understand how to use the virtual room targets. My approach has been to edit the harman-room-target.csv file. A bit arduous but doable. But I've had mixed results trying to use it to adjust the final freq response curve. Basically where I'm at now is leaving the Harmon adjustments below 20 and above 12k and setting it flat in between. I can more easily adjust eq at the end using Peace and hear the results in real time. It's not nearly as granular as the Harmon file but as you say bass/mid/treble adjustment may be sufficient.

In some ways having a HRIR means we can do A vs B so much quicker and easier than on a real loud speaker setup. But in other ways because you can switch so quickly between HRIR's you don't give your brain enough time to compensate. You then end up fiddling with many settings. For my own listening i've created 3 go-to HRIR's. One is nearfield that I use when close to a screen - like a laptop or desktop monitor, the other is midfield where I'm sitting 1-2m away from a TV and the final is far-field whic his for projection use. The only one I can consistently use in all situations is nearfield - if I use the far field one in a different environment it just sounds wrong.

Switching between HRIRs quickly is definitely a bit disconcerting. Some sound very good at first with certain tracks, but serious flaws can show up on other tracks. And there is a period of hearing adaptation to a different tonal balance and imaging presentation.

I'm currently getting good repeatabliity with my headphone measurements. What I believe is the biggest cause of the variablity are the room measurements. Some of my HRIR's are very tight, the imaging is so precise, and others are much more spread out, with a much broader spacious sound stage but vague blurred imaging. Voices vary greatly from really excellent to echoy as if in a cave. I'd like to be able to pick and choose attributes or fix specific elements, but figuring out what the cause is and how to make surgical adjustments for me is very elusive. In my fantasies there are sliders or knobs to make changes in key parameters, like eq now, while listening, but that could be beyond what's feasible. As you say it's some of my audio-OCD. But the results can be so good, it's such a breakthrough, it's hard not to want to perfect it when something odd shows up on a track. I could very well end up with several HRIRs as you have, switching for the type of music. Pretty nice problem to have where you can make that level of sound system change with the click of a mouse.

 

[jaakkopasanen]

I redid the Harman room target because the old one had too much bass. I don't know where I did get the previous one but the new one has about 6.8 dB of bass boost. This new target is called harman-in-room-loudspeaker-target. I also added another target which Harman's headphone target for in room setup. It's essentially the same but with less treble. This is called harman-in-room-headphone-target.

Another update is the bass_boost and tilt parameters which work in the same way as in AutoEQ. Now it's possible to use Harman room targets without bass and add desired level of bass boost with the parameter. Tilt makes the whole frequency response darker or brighter.

 

[lowdown]

Just want to say Impulcifer is so amazingly good. I thank you every day.

That is all.

 

[jaakkopasanen]

Just want to say Impulcifer is so amazingly good. I thank you every day.

That is all.

Aww. Thanks!

 

[musicreo]

I also added another target which Harman's headphone target for in room setup. It's essentially the same but with less treble. This is called harman-in-room-headphone-target.

If I have a headphone measurement how do I actually apply the targets? What is the command for that?

By default the headphone compensation is invidual for the left and right channel or is a average of both channels used? If I only want to use left or the right measured channel for the headphone compensation how do I do this without editing the headphones.wav?

Attached is one of my latest headphone maesurements (HD555 measured with Primo EM258). The difference in the frequency response begin already at 2 kHz. I guess invidual compensation in this "relativ low" frequency range is problematic as the mic placement may have a different effect on the speaker measurement compared to the heaphone measurement?

 

[johnn29]

Been loving my HRIR for sometime now. I'm spending another month away from home in an airbnb. Still can't get over how with a 15.6" 4k OLED and Impulcifer I can be transported back to my theater room - had a great experience watching Midway on the laptop. Surround track on that is excellent!

Just to chime in on my thesis that doing A vs B very quickly on HRIR isn't a great way to test and that you need multiple HRIR's for multiple physical rooms/situations so you get used to the sound of the room. Sean Olive (researcher at Harman) cites some research that may be of interest that seems so support this: https://www.stereophile.com/content/slow-listening

 

[venkatesh63]

@jaakkopasanen @johnn29 @musicreo can you share your recordings of HRIR.i don't want demo recordings from Impulcifer page. i want your created copies of different speaker's and Equalization's(eq.csv) of headphone's compensation files. i want for Movies and Music to listen copies of hesuvi.wav or hrir.wav you guys created with your gears.