• 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!

How much do the ears really smooth

bennybbbx

Addicted to Fun and Learning
Joined
May 27, 2020
Messages
695
Likes
126
Location
germany
I do a test 2 measure with both speakers on at same time and 0 delay. i use stative and the 2. measure confirm that measure is very simular at 1/2 octave smooth. mix position and speaker position stay in all 4 measure
right speaker delay 5 cm step response.jpg
mtm both delay 0 step response.jpg
both, right delay 5 cm.jpg

both all delay 0 frequency.jpg
same. Now i delay the right speaker with 5 cm(0,05 m) in the t racks mini dsp and do 2 measures again. look at frequency responese they are very diffrent but what hear is only little diffrent when play mono song. i upload also all the measure so you can look yourself. when do more smoothing frequency diffrence is not so much and come more near what hear.

here can download the mdat.

https://www.file-upload.net/download-14443254/mtmwithandwithoutdelay.mdat.html
with delay less smooth.jpg
no delay less smooth.jpg
1_1 smooth no delay.jpg
right delay 1_1 smooth.jpg


1/1 smooth is good for 10 band graphic EQ. more than 1 octave smooth is in REW not possible
 

Attachments

  • both 0 step response show it too.jpg
    both 0 step response show it too.jpg
    760.9 KB · Views: 133
Last edited:
bennybbbx.png

This why you see a huge diff in the magnitude response while not hearing too much difference:

In the first measurement, the single mic sees a pulse doublet as the distance to both speakers is not the exctly the same to begin with. In the second measurement (blue) again we see a doublet from your added delay, with a different spacing, this time. In the FR this manifests as comb-filtering and the dips land at different frequencies depending on the spacing.

When you listen, you do with two "mics" which dominantly pickup the signal from the speaker on the same side so the comb filtering is much less.
While the difference can easily be heard with pink noise (including the associated image shift to the non-delayed side), with normal music signal it is much less noticable unless you pay close attention to the high frequency range.
 
Not sure I understand the question. Are you asking which smoothing of response is closest to how our ears hear things? If so the answer is it varies with frequency. At higher frequencies 1/4 octave and 1/5th octave smoothing is near what our hearing does. Middle frequencies are closer to 1/3 octave and the lowest maybe 1/2 octave.

What describes our ear's smoothing is ERB, or effective rectangular bandwidth. This describes what frequency band our hearing lumps together at any given center frequency in terms of masking. ERB is one of the smoothing choices in REW. Another is psycho-acoustic which includes ERB and some more advanced adjustments. I find psychoacoustic to come closest to what my ears hear like.
 
View attachment 106684
This why you see a huge diff in the magnitude response while not hearing too much difference:

In the first measurement, the single mic sees a pulse doublet as the distance to both speakers is not the exctly the same to begin with. In the second measurement (blue) again we see a doublet from your added delay, with a different spacing, this time. In the FR this manifests as comb-filtering and the dips land at different frequencies depending on the spacing.

When you listen, you do with two "mics" which dominantly pickup the signal from the speaker on the same side so the comb filtering is much less.
While the difference can easily be heard with pink noise (including the associated image shift to the non-delayed side), with normal music signal it is much less noticable unless you pay close attention to the high frequency range.

that the blue have diffrent distance too between left and right speakers i did not see before. I look closer and see this too. so you think it is much diffrence if hear with 2 ears instead 1 microphone ?. maybe i need a head with microphone inside thats simulate better the hide from opposite site. as far can see the freq changes happen above 1 khz.

did somebody know a cheap head with microphones inside that is used for binaural recording.

That ears do smoothing at higher freqeuncy less i read too. but i think 1/ 4 - 1/6 is too few. or maybe ears do a additional stereo smoothing and smooth also left and right side. but to see more, i maybe should do a example with EQ. use 4 khz Q 2 with + 6 db and then 6 khz Q2 -6 db. if the diffrence is much or less hearable.
 
you think it is much diffrence if hear with 2 ears instead 1 microphone ?

Play 1kHz tone with both speakers, and move your head around slowly by at least a couple of feet. You might hear the tone dim a little.

Now, plug one ear (simulate one microphone) and repeat. You may find the tone becomes inaudible in some locations.
 
When you listen, you do with two "mics" which dominantly pickup the signal from the speaker on the same side so the comb filtering is much less.
While the difference can easily be heard with pink noise (including the associated image shift to the non-delayed side), with normal music signal it is much less noticable unless you pay close attention to the high frequency range.


I get now the idea to hear music more as the mic and i rotate my head 90 degree. so i hear only with 1 ear. in the other ear i put an ear plug. then i hear music and shift the delay of right speaker. I hear around 2-3 db frequency change but not 6 db and more as the mag show at 1/2 smoothing.

the question also is when audio science say psychoacoustic is as ears hear. how did they measure it and confirm that it is right.
 
Play 1kHz tone with both speakers, and move your head around slowly by at least a couple of feet. You might hear the tone dim a little.

Now, plug one ear (simulate one microphone) and repeat. You may find the tone becomes inaudible in some locations.

with sinus signals i hear extrem much diffrence in compare to music. when i play a 5 khz sine tone with both speakers and my head rotate 90 degree, so only 1 ear point to speaker then i hear really no testtone when i delay the right speaker to 4,3 cm. maybe the ERB is create with hear sine tones, but this is not what do with speakers and hear musik. i think when hear music ear smoothing is much more as with sine tones and the ERB or psychoacoustic smoothing theory is only correct when hear sine tones
 
Not sure I understand the question. Are you asking which smoothing of response is closest to how our ears hear things? If so the answer is it varies with frequency. At higher frequencies 1/4 octave and 1/5th octave smoothing is near what our hearing does. Middle frequencies are closer to 1/3 octave and the lowest maybe 1/2 octave.

What describes our ear's smoothing is ERB, or effective rectangular bandwidth. This describes what frequency band our hearing lumps together at any given center frequency in terms of masking. ERB is one of the smoothing choices in REW. Another is psycho-acoustic which includes ERB and some more advanced adjustments. I find psychoacoustic to come closest to what my ears hear like.

Is it advisable to do EQ based on less smoothing, even if our ears can't discriminate that well? Typically, I've used 1/12 or 1/24 for EQ purposes, but maybe I'm overcorrecting? If I set it to ERB/Psycho or 1/3, it almost makes it look like there is nothing to EQ(other than overall tilt), whereas subjectively, I know my 1/12 or 1/24 based PEQs have improved the sound, though I suppose that's easily explained by expectation bias :(
 
when do more EQ you get more stronger phase shifts diffrences. I do the next days a example with mfreeform eq match EQ and diffrent smoothing values. when use small smoothing it use small bands and much EQ levels it sound a little as flanger metallic and crisp. there is a large range that hear no diffrence. but if your ears did hear or not need test.
 
here is a video that use 1/3 bandwidth and as often happen when measure with few smoothing, after a large peak follow immidiate a large gap. this i simulate from a measure with the EQ there can hear phasing effects on

that correction with 1/12 sound better as without must not be expection bias. in can sound better, but with less EQ settings can even more sound better. or more realistic
 
The crosstalk comb filtering isn't as bad sounding as one would expect from looking at measurements. The brain does smooth it out by using both ears at once. But still, it's really nice when it's gone, such as when using crosstalk elimination methods.
 
I think the erb and psychoacoustic smoothing are only valid for noises.
note B8 is at 7902Hz. if you have a thin notch of 10dB there you will obviously hear the lack of volume when this note plays. it will be smoothed out in erb scale though
 
I think the erb and psychoacoustic smoothing are only valid for noises.
note B8 is at 7902Hz. if you have a thin notch of 10dB there you will obviously hear the lack of volume when this note plays. it will be smoothed out in erb scale though

you can hear with a phase linear freeform EQ a new example A B. if you hear diffrence. C have very less smoothing it sound bad.I hear no diffrence between A and B(or a slight feeling that the room in B sound better. only C sound clear bad.


or what do you think ?
 
Last edited:
I have edit, now the link should work
 
I think the erb and psychoacoustic smoothing are only valid for noises.
note B8 is at 7902Hz. if you have a thin notch of 10dB there you will obviously hear the lack of volume when this note plays. it will be smoothed out in erb scale though

in music also synthesizer use always at least 2 oscilators(which produce only simular frequency because they are a little detune). music with only 1 frequency sound not good for most. to see how it sound without vibrato you can hear heavy autotune vocals. the harmonics are there but the frequency movement and detune is delete. or use a piano guitar trumpet or what you like and try loop 1 wave cycle of it. it sound worse
 
you can hear with a phase linear freeform EQ a new example A B. if you hear diffrence. C have very less smoothing it sound bad.I hear no diffrence between A and B(or a slight feeling that the room in B sound better. only C sound clear bad.


or what do you think ?

sorry for late reply. hard to say if I hear the diference of a and b, I think I do, but have to blind test it.

also my point was that you would here it on specific notes (in that case a 120Hz-ish note). I am not saying that notes only have one frequency, but try out EQing a isolated instument in a DAW and find out that a few dB's in energetic frequencies will have a big impact
 
sorry for late reply. hard to say if I hear the diference of a and b, I think I do, but have to blind test it.

also my point was that you would here it on specific notes (in that case a 120Hz-ish note). I am not saying that notes only have one frequency, but try out EQing a isolated instument in a DAW and find out that a few dB's in energetic frequencies will have a big impact

see the example video in the attached file. it is a high note at f6 that use a sound that sound simular to a simple wave. but can see, it use much neighbour frequencies
 

Attachments

  • lead sound.zip
    630.3 KB · Views: 120
see the example video in the attached file. it is a high note at f6 that use a sound that sound simular to a simple wave. but can see, it use much neighbour frequencies

I see where you are going, but it is not as easy as that. analysers use windowing. they will show neighbor frequencies even on sine notes
 
I see where you are going, but it is not as easy as that. analysers use windowing. they will show neighbor frequencies even on sine notes

yes thats the question how much "windowing" ears use. i did not know how much windowing the analyser use. it have many settings. this are default settings i use. you can see in the lower frequencies of the root frequency it go much faster to lower levels. all analyser i see so far use this settings. I can change the settings overlap to 1 or rectangle window curves look very simular(have no large gaps between them). EDIT: I notice now the smoothing for the correction EQ is also use for the analyizer.in the audio test i do it is 5%. when think 5% sound not better(in compare to linear headphones) as 1% then can hear that much smoothing is not good. when use minimum phase few smoothing in the EQ correction sound even more worse.
analyzer settings.jpg
 
Last edited:
Back
Top Bottom