MrHaelscheir
500+ Head-Fier
We are certainly not driving 48V directly into our ears. The VXLR+ adapters bring that down to around 5V "plug-in-power" for the microphone capsules in question.
Recording Impulse Responses for Speaker Virtualization
- Thread starter jaakkopasanen
- Start date
Hello, I have some interesting news to share. Has everyone heard about the new version of VB-Cable? Previously, VB-Cable was limited to 2-channel output (in 7.1 / out 2). With the release of VB-Matrix, it was expanded to 8-in and 8-out, and someone on the VB forum requested an increase in channel capacity. About a month ago, a new version was released. Now it supports 16-in and 16-out!
In APO, it doesn't recognize height channels like "Lheight"; instead, they're simply assigned as channels 1 through 16. Additionally, without a receiver, decoding is not possible, so true Atmos playback still seems challenging on Windows. However, it does open up a lot of possibilities. For instance, it’s now feasible to implement upmixing with added height channels. Plus, 7.1ch (8ch) can now be handled equally on both input and output, with full 8-channel support.
I'm not sure; I haven't looked deeper into other possible ways to decode Atmos on Windows. But I hope this information is helpful.
In APO, it doesn't recognize height channels like "Lheight"; instead, they're simply assigned as channels 1 through 16. Additionally, without a receiver, decoding is not possible, so true Atmos playback still seems challenging on Windows. However, it does open up a lot of possibilities. For instance, it’s now feasible to implement upmixing with added height channels. Plus, 7.1ch (8ch) can now be handled equally on both input and output, with full 8-channel support.
I'm not sure; I haven't looked deeper into other possible ways to decode Atmos on Windows. But I hope this information is helpful.
musicreo
100+ Head-Fier
- Joined
- Jun 25, 2011
- Posts
- 401
- Likes
- 152
hmm... interesting.
It seems that you can only convert files you have, not streaming platforms like Netflix.
But it still looks interesting.
Kosorm
New Head-Fier
I saw someone have decent luck using foam eartips with their iebms. Can confirm atleast for the SP-TFB-2s (the cheaper of the sound professionals models) it works wonderfully, you just have to cut out the casing besides the part on the back that protects the electronics. no glue required!
Last edited:
Kosorm
New Head-Fier
Hi again. This post is going to have a few questions and reference a guide i found on reddit (and is one of the first results on google for impulcifer), if this is the wrong place to ask ill delete the post sorry.
I've been following this guide as well as reading as much of this thread as possible and the measurements page on the wiki, and while trying to follow the guide, ive run into a few problems that im unsure how to go about fixing. Ive never used REW before this project but have been able to mostly figure things out by trial and error or googling. Below are the steps laid out in the guide that ive had some trouble with.
Step 1. Iemb calibration
Is this step even necessary? I followed it without issue and i can sort of understand the logic of it, but the fact i havent been able to find anyone else suggesting doing this (maybe i used the wrong search terms) has me wondering about its inclusion in the guide.
Step 2. Speaker calibration
Same issue as the first although ive looked much less into this one. On top of that im unsure of how to complete this step as a whole. Ive gathered that i need to have rew generate pink noise and go into the rta menu but from there im sort of puzzled. I can input most of the settings as guided, but im not sure how long to do the measurement, how to import the house curve (am i supposed to do it in the eq menu for this?), and then when matching the eq to the curve, am i supposed to do anything manually here or just use the function to match it automatically?
Step 3+4 are simple enough.
Step 5. EQing your headphone
Same question as before, do i have to manually make the eq for it or can i just let the program do its thing? Im not super adept with making eq adjustments so im guessing if the latter is the case i should probably learn that?
Step 6 and beyond seems documented well enough but i am not at that point yet so i have no questions at this time.
Thanks sorry if this was a bit of a long post, I really appreciate any help i can get with this.
I've been following this guide as well as reading as much of this thread as possible and the measurements page on the wiki, and while trying to follow the guide, ive run into a few problems that im unsure how to go about fixing. Ive never used REW before this project but have been able to mostly figure things out by trial and error or googling. Below are the steps laid out in the guide that ive had some trouble with.
Step 1. Iemb calibration
Is this step even necessary? I followed it without issue and i can sort of understand the logic of it, but the fact i havent been able to find anyone else suggesting doing this (maybe i used the wrong search terms) has me wondering about its inclusion in the guide.
Step 2. Speaker calibration
Same issue as the first although ive looked much less into this one. On top of that im unsure of how to complete this step as a whole. Ive gathered that i need to have rew generate pink noise and go into the rta menu but from there im sort of puzzled. I can input most of the settings as guided, but im not sure how long to do the measurement, how to import the house curve (am i supposed to do it in the eq menu for this?), and then when matching the eq to the curve, am i supposed to do anything manually here or just use the function to match it automatically?
Step 3+4 are simple enough.
Step 5. EQing your headphone
Same question as before, do i have to manually make the eq for it or can i just let the program do its thing? Im not super adept with making eq adjustments so im guessing if the latter is the case i should probably learn that?
Step 6 and beyond seems documented well enough but i am not at that point yet so i have no questions at this time.
Thanks sorry if this was a bit of a long post, I really appreciate any help i can get with this.
sander99
Headphoneus Supremus
@Kosorm: I don't know those guides, what about the following, maybe these can help you:
https://github.com/jaakkopasanen/Impulcifer
https://github.com/jaakkopasanen/Impulcifer/wiki/Measurements
https://github.com/jaakkopasanen/Impulcifer
https://github.com/jaakkopasanen/Impulcifer/wiki/Measurements
Kosorm
New Head-Fier
Appreciated. As i said above though i am also following the wiki and this thread in addition to the guide
I will respond to each point in your message.
It may or may not be necessary.
If the angle and insertion depth are exactly the same, it might not matter, but even then, the acoustic characteristics of both ears can differ, which means the ultra-high-frequency response might inherently vary between ears.
Now, even without a calibration curve, let’s assume the following:
For instance, let’s say the left (L) microphone has a flat response (zero), and the right (R) microphone has a response as if a negative high-shelf filter is applied.
In this case, the same characteristic will already be applied to your headphone profile and room response, for each channel and for each input microphone.
The L microphone was flat—no change.
As for the R microphone, we assumed it has a response with a negative high-shelf filter.
Let’s further assume that the room response applies a roll-off in the ultra-high frequencies, resembling a negative high-shelf filter.
In the headphone response, the same negative high-shelf filter would apply.
Since the headphone response would overcompensate by the exact amount of the negative high-shelf applied, the end result wouldn’t show a significant difference.
Therefore, you can skip the calibration step.
You don’t need to place too much importance on this.
It doesn’t matter whether you use pink noise with an RTA or a sine sweep. Measuring without applying EQ first and then applying EQ afterward will yield the same result.
In fact, I argue that the use of DRC systems like Dirac, which employ FIR filters or mixed-phase processing, is better avoided. This is a view I advocate within the Korean community I’m active in.
Therefore, if there’s anything you need to adjust, simply apply basic EQ to the low bass range (20–200 Hz) and leave it at that.
As these become smoothed out, the result drifts further away from your natural listening perception, making it less precise.
I’ve experimented not only with basic smoothing but also with techniques like FDW (Frequency-Dependent Windowing) and MTW (Multi-Time Windowing).
Ultimately, the most accurate representation was the raw, unsmoothed state.
Using PEQ to smooth out room modes doesn’t require many filters and is straightforward.
If you know how to use REW, you can also use its AutoEQ function.
It may or may not be necessary.
If the angle and insertion depth are exactly the same, it might not matter, but even then, the acoustic characteristics of both ears can differ, which means the ultra-high-frequency response might inherently vary between ears.
Now, even without a calibration curve, let’s assume the following:
For instance, let’s say the left (L) microphone has a flat response (zero), and the right (R) microphone has a response as if a negative high-shelf filter is applied.
In this case, the same characteristic will already be applied to your headphone profile and room response, for each channel and for each input microphone.
The L microphone was flat—no change.
As for the R microphone, we assumed it has a response with a negative high-shelf filter.
Let’s further assume that the room response applies a roll-off in the ultra-high frequencies, resembling a negative high-shelf filter.
In the headphone response, the same negative high-shelf filter would apply.
Since the headphone response would overcompensate by the exact amount of the negative high-shelf applied, the end result wouldn’t show a significant difference.
Therefore, you can skip the calibration step.
You don’t need to place too much importance on this.
It doesn’t matter whether you use pink noise with an RTA or a sine sweep. Measuring without applying EQ first and then applying EQ afterward will yield the same result.
In fact, I argue that the use of DRC systems like Dirac, which employ FIR filters or mixed-phase processing, is better avoided. This is a view I advocate within the Korean community I’m active in.
- Impulcifer’s impulse detection and processing are highly sensitive.
- As mentioned in point 1, we cannot know whether you wore the microphones correctly. If the response is inconsistent or if the microphone faces the wall of the ear canal, the resulting impulse will naturally differ.
- We also don’t know if your speakers are linear or if direct sound from your speakers is free of interference in your room when it starts.
Therefore, if there’s anything you need to adjust, simply apply basic EQ to the low bass range (20–200 Hz) and leave it at that.
The numerous dips and peaks within the ear reflect the way you perceive sound.
As these become smoothed out, the result drifts further away from your natural listening perception, making it less precise.
I’ve experimented not only with basic smoothing but also with techniques like FDW (Frequency-Dependent Windowing) and MTW (Multi-Time Windowing).
Ultimately, the most accurate representation was the raw, unsmoothed state.
You can apply it manually; it’s not difficult.
Using PEQ to smooth out room modes doesn’t require many filters and is straightforward.
If you know how to use REW, you can also use its AutoEQ function.
Kosorm
New Head-Fier
Hey thank you so much for the response, this is super helpful information and will be kept in mind tomorrow when i do more measurements
Last edited:
Kosorm
New Head-Fier
Getting this error, I think my iebms were plugged into my audio interface with left in right and right in left by accident. is there any way i can flip the two channels so i dont have to remeasure?
C:\Users\orpho\Impulcifer\hrir.py:223: UserWarning: Warning: FL measurement has lower delay to right ear than to left ear. FL should be at the left side of the head so the sound should arrive first in the left ear. This is usually a problem with the measurement process or the speaker order given is not correct. Detected delay difference is 749.8125 milliseconds.
warnings.warn(f'Warning: {speaker} measurement has lower delay to right ear than to left ear. '
C:\Users\orpho\Impulcifer\hrir.py:209: UserWarning: Warning: FR measurement has lower delay to left ear than to right ear. FR should be at the right side of the head so the sound should arrive first in the right ear. This is usually a problem with the measurement process or the speaker order given is not correct. Detected delay difference is 0.3333 milliseconds.
warnings.warn(f'Warning: {speaker} measurement has lower delay to left ear than to right ear. '
Edit: I may have fixed this(?) but i have another question. How do i use the HRIR if i just want to do stereo and not surround? I have hesuvi up but it seems to only want to do surround sound and the impulcifier github doesnt really say anything about this
C:\Users\orpho\Impulcifer\hrir.py:223: UserWarning: Warning: FL measurement has lower delay to right ear than to left ear. FL should be at the left side of the head so the sound should arrive first in the left ear. This is usually a problem with the measurement process or the speaker order given is not correct. Detected delay difference is 749.8125 milliseconds.
warnings.warn(f'Warning: {speaker} measurement has lower delay to right ear than to left ear. '
C:\Users\orpho\Impulcifer\hrir.py:209: UserWarning: Warning: FR measurement has lower delay to left ear than to right ear. FR should be at the right side of the head so the sound should arrive first in the right ear. This is usually a problem with the measurement process or the speaker order given is not correct. Detected delay difference is 0.3333 milliseconds.
warnings.warn(f'Warning: {speaker} measurement has lower delay to left ear than to right ear. '
Edit: I may have fixed this(?) but i have another question. How do i use the HRIR if i just want to do stereo and not surround? I have hesuvi up but it seems to only want to do surround sound and the impulcifier github doesnt really say anything about this
Last edited:
Kosorm
New Head-Fier
Still might be doing it wrong but i tried throwing in the hrir.wav into eapo through the convolution with impulse response after manually swapping the left and right channels in both fr and fl.wav and for some reason i keep getting a loud white noise type hiss whenever audio is playing on my pc. this is my file anyone know whats happening? my hrir.wav
the hissing is entirely in my left ear
the hissing is entirely in my left ear
Last edited:
sander99
Headphoneus Supremus
If I remember all this correctly:
You should have the 'Upmix Content' options unchecked, or at least the stereo option under 'Upmix Content' unchecked. Otherwise it will upmix stereo sources to surround.
When HeSuVi is in 5.1 or 7.1 mode and you play a stereo source while there is no upmixing then just the virtual front left and right speakers will play that stereo content and the other virtual speakers will be silent, so effectively it doesn't matter that you are not in stereo mode.
I've seen your writing now.
[Looking at the comments] just a momment
You use Audacity?
Can you upload FL,FR.wav file? (original no edit)
You don't have to do anything.
If you only have FL,FR, you will mute the rest of the channel.
also why you use HRIR.wav?
Are you writing hisuvi?
If so, you need to use Hesuvi.wav for Hesuvi. (HRIR.wav and Hesuvi.wav are the same, but the track order is different.)
Last edited:
Users who are viewing this thread
