• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

example binaural files of a) normal room response and b) time aligned / exess group delay corrected

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,675
Location
Salvador-Bahia-Brasil
So I have been working on some binaural impulse responses. My goal was to frequency correct them. but I ended up with 2 versions:
x: frequency corrected in REW
y: phase corrected with Denis Sbragion's DRC and then with REW afterwards (because DRC doesn't result into totaly flat)

they are pretty close (var smoothing). there are obviously slight diferences in the no smoothing plots

fr.jpg


and here are the step responses and exess group delays (yes, the corrected one has visable pre-delay. but will it sound better or worse?)

step1.jpg
excess1.jpg
setp2.jpg
exc2.jpg


now the coolest thing about this is that you guys will be able to hear the diference. I created some small files which switch between the two convolutions.
So now grab your headphone!
all these files start with convolution x, and then switch to y, and back and forth.
pay attention to the kick drum manly. and also don't forget that the kick has 2 sounds (atack and decay)

2 loops with x, then 2 loops with y
https://drive.google.com/file/d/1feMtu_EmmBkIqo8Z7gDWMK7LlDuaf64Z/view?usp=sharing

8 loops with x, then 8 loops with y, then 8 loops with x, then 8 loops with y
https://drive.google.com/file/d/1mt2qaRKPyKpi9f7tR30UZouP93PVA6iG/view?usp=sharing

2 loops with x, then 2 loops with y
https://drive.google.com/file/d/1CZcNZVI2fwNqELj8RMw5RZl7Loy_GmxM/view?usp=sharing

x, y, x, y
https://drive.google.com/file/d/1HYd5bHXj5A9PJD8bDmYG9zlDToRcsUpx/view?usp=sharing

what about a song?
x, y, x, y
https://drive.google.com/file/d/1fknMD7urvi9qHj4Op5kwLXVClPMh2t6M/view?usp=sharing

I wont coment on any of those. want to hear you guys first

EDIT: if you guys want some additional examples feel free to post a link of a sound from https://freesound.org
thinking about maybe doing bass examples.
I just heard this electronic kick and the diference is ridiculous https://freesound.org/people/red19official/sounds/436506/
 
Last edited:
OP
dasdoing

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,675
Location
Salvador-Bahia-Brasil
well, I didn't necessary expected positive feedback, but 0 feedback?
however, I tested a time aligned version (diferent IR) exessivly today and I never heard the pre-delay.
thinking about it, in this specific case. the choice is between what is that, a 12ms pos-delay peak with 93% energy, and a 2ms pre-delay peak with 31% energy. yes, the pos delay will be masked, but the audio semples clearly showed me that the corrected version is preferable. Audio example number 3 is a clear wink. the uncorrected version doesn't even sound natural.
I will aply this soon on my real speakers. it's possible it wont work out like this in the real world, but in theory the result should be the same at listening position at least
 

Berwhale

Major Contributor
Forum Donor
Joined
Aug 29, 2019
Messages
3,935
Likes
4,925
Location
UK
well, I didn't necessary expected positive feedback, but 0 feedback?

Maybe add voting buttons? :)

I preferred the sound of Y each time, but I suspect this was because I was listening more intently to Y to see if I could find differences to X. i.e. I may have picked X if the responses had been reversed.
 
OP
dasdoing

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,675
Location
Salvador-Bahia-Brasil
Maybe add voting buttons? :)

I preferred the sound of Y each time, but I suspect this was because I was listening more intently to Y to see if I could find differences to X. i.e. I may have picked X if the responses had been reversed.

but x and y reversed. i thought this way it would be possible to compare. but thanks for feedback
 

bogart

Active Member
Forum Donor
Joined
Sep 10, 2019
Messages
248
Likes
304
edit: I got them all to play. Listened on slightly bass-shy in ear monitors.

Track 1: noticeably better attack then sustain/decay from y. I think I could ABX this difference at 100% with preference for y.

Track 2: I heard the difference here best in the high tom, it has a noticeable ringing in the corrected, vs. more of a splatty attack with a ring in x. This one is more subtle to my ears.

Track 3: a little subtle also but became clearer on repeat as my ears trained to the sound better. I was definitely able to guess the repeat pattern without referencing your note.

Track 4: my notes on this track were that the bass drum sounds lower on y. I suspect that is because the drum hit is at higher frequency and the resonance is lower, by aligning better I’m able to hear the deeper tone in the resonance?

Track 5: if you’ve ever played with electronic drum sounds, x would be “wet” and y would be “dry” to my ear.

Overall, I think you definitely helped me train my ears better to find the issue and I can appreciate the difference from the extra step. Thanks for posting these.
 
Last edited:
OP
dasdoing

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,675
Location
Salvador-Bahia-Brasil
I’ve only been able to play the first clip on my phone, but it does have a noticeably more correct sound in version y. The attack from the hit gets out of the way and the sustain/decay follows properly.

I will say it helped that I knew what difference to listen for and which was which. However, there’s no doubt they sound different, and I’m fairly sure I could ABX this difference near 100% :)

thanks
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
I'd prefer if you had separate files for x and y so I could use it in an ABX comparater.

I did listen over speakers. Didn't hear any difference in the 1st one or 2nd one.

The 3rd one I think I do, but the way you have switched between them seems timed differently which may be what I'm hearing. Again a couple files one with X and Y would be easier to use.

4th one I don't think I hear, though it also seems the time between the beat and switching sources differs which clouds whether I'm hearing that or something else.

On the song, I think I hear the shuffle shaker sound as being rather different. But having them combined this way I think hampers how easy it is to hear.
 
OP
dasdoing

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,675
Location
Salvador-Bahia-Brasil
I don't think ABX is a good tool in this case. ABX is more suited for cases like "do I hear the diference?". I think diferentiating the two in this case is easy.
also @Blumlein 88, you realy need to hear these on headphones as you speakers will have aditional group delay justly in the region we are testing here.

edit: I got them all to play. Listened on slightly bass-shy in ear monitors.

Track 1: noticeably better attack then sustain/decay from y. I think I could ABX this difference at 100% with preference for y.

Track 2: I heard the difference here best in the high tom, it has a noticeable ringing in the corrected, vs. more of a splatty attack with a ring in x. This one is more subtle to my ears.

Track 3: a little subtle also but became clearer on repeat as my ears trained to the sound better. I was definitely able to guess the repeat pattern without referencing your note.

Track 4: my notes on this track were that the bass drum sounds lower on y. I suspect that is because the drum hit is at higher frequency and the resonance is lower, by aligning better I’m able to hear the deeper tone in the resonance?

Track 5: if you’ve ever played with electronic drum sounds, x would be “wet” and y would be “dry” to my ear.

Overall, I think you definitely helped me train my ears better to find the issue and I can appreciate the difference from the extra step. Thanks for posting these.

my impression on track 1 and 3: x doesn't realy sound bad, then y comes and it is a aha moment. y sounds so much more natural
track 2 is the only one where I actualy have dificulty on deciding which I prefer. maybe because I don't realy like how the fill sounds.
in 4 and 5 x lacks kick drum "punchyness", which y constructs
 

bogart

Active Member
Forum Donor
Joined
Sep 10, 2019
Messages
248
Likes
304
I don't think ABX is a good tool in this case. ABX is more suited for cases like "do I hear the diference?". I think diferentiating the two in this case is easy.
also @Blumlein 88, you realy need to hear these on headphones as you speakers will have aditional group delay justly in the region we are testing here.
I think ABX is useful, though. If you’re debating, for example, a miniDSP DDRC to improve your room correction game, it’s worth trying to see whether the correction is audible to you, or whether frequency correction alone meets your needs. It’s also just practically difficult here with the abrupt transitions between the loops being a bit jarring to the listener.

In my case, I played drums long enough that these corrections hit a sweet spot for me to hear right and wrong. I’m not sure they’d be as obvious if my ear weren’t trained on the source material, as @Blumlein 88 points out. I’d be able to tell for certain with an ABX, particularly on program material instead of samples.
 
OP
dasdoing

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,675
Location
Salvador-Bahia-Brasil
you guys want the clips seperated? I would just cut them in my DAW. not much work

very nice to have a drums player to check this out
 
OP
dasdoing

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,675
Location
Salvador-Bahia-Brasil
OK, I finaly tested it on my speakers and it has the same result. I will definatly stick to the step response correction at listening position.
btw, I use Align 2 as a front end for Denis Sbragion's DRC, which is a freeware software and which was intended for Minidsp in version 2: https://www.minidsp.com/applications/advanced-tools/align2-for-drc
you will find zero about it in the internet and I suspect it is because it seams so complicated, when it realy isn't.
Denis Sbragion's DRC is very underated; it's free and probably does way more then all the expensive foolproof solutions for room correction.
if there is interest I can make a guide for how I use it in a seperate topic.
 
Last edited:

Propheticus

Senior Member
Joined
Nov 18, 2020
Messages
431
Likes
644
Location
Vleuten, Netherlands
Well, that was hard. Hardly hear any difference, apart from the bass where X sounds 'wet' and y 'dry'. The change is minimal.
So minimal I started wondering if you were secretly testing expectation bias and x=y. A look at the waveforms told me that's not the case.
It does also clearly show the ~10ms shift of 'pre-ring'. I wonder if the 10ms shift of impulses causes any interference at 100Hz and it's multiples. Might explain the boost in bass you describe.
On the song (sample 5) I actually prefer X (= original?), Y has a slight metallic/unnatural sound to it.
 
OP
dasdoing

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,675
Location
Salvador-Bahia-Brasil
Well, that was hard. Hardly hear any difference, apart from the bass where X sounds 'wet' and y 'dry'. The change is minimal.
So minimal I started wondering if you were secretly testing expectation bias and x=y. A look at the waveforms told me that's not the case.
It does also clearly show the ~10ms shift of 'pre-ring'. I wonder if the 10ms shift of impulses causes any interference at 100Hz and it's multiples. Might explain the boost in bass you describe.
On the song (sample 5) I actually prefer X (= original?), Y has a slight metallic/unnatural sound to it.

I have not concluded my studie on the subject yet,
I will respond to it in this newer topic. feel free to test those files, too https://www.audiosciencereview.com/...ion-on-binaural-room-impulse-responses.18694/
 
Top Bottom