EarSpace!!!!

[esldude]

5 hours ago, Don Hills said:

 

The room is the elephant when it comes to recording a system in this way.

Outside is much better:

 

 

 

 

 

 

 

The stadium seemed to corrupt the sound recorded more than the other two. 

[esldude]

I made some recordings of music for FAS42 once.  I recorded speakers in my room using cross-pairs of mics with different pickup patterns at the listening position.  You get an echo filled distant sound.  The reason is early room reflections.  You also can get uneven bass if you pick a poor spot in the room.  The reason is our ears of course respond to those same reflections, but our brain filters them out for the first 10 milliseconds or so.  We "hear thru the room" in that sense to mainly the direct speaker sound.  Reflections picked up by microphones get reproduced as direct reproductions of those reflections and we hear them.  That is a problem with binaural or any other miking at the listening position.  

 

Now I also did one of the recordings with a mic 1.5 meters from each speaker. That one sounded very much like the speaker sounds at the listening position because the direct to reflected sound ratio was much more in favor of direct sound.  Due to our hearing filtering out reflections we hear a much more direct sound from speakers.  

 

And that is the point of Don's video links.  Outdoor recordings of decent PA speakers sound less colored and more like the music because there are no reflections compared to recording speakers in a normal listening room at the listening position.  That is also why the stadium sounded less good than the other two.  

[esldude]

6 hours ago, pkane2001 said:

 

Same question I asked Don: why doesn't the brain filter out reflections in a recording? What's different? Are reflections not recorded correctly? Or is it that your visual perception prevents it?

 

I believe you said before that binaural recordings don't sound good to you. They sound amazing to me, so there may be some differences in how we process sound or possibly differences in our ear/head anatomy that affect sound.

 

A while back, I decided to see if I can reproduce the effect of my room through headphones. I used REW to create an impulse response from the measurements collected at my normal listening position. I then used this to convolve with the music played back through headphones. It sounded great! The effect was an added dimension to the sound, with a bit more depth and sense of space and reverb but no unwelcome distortions. Obviously, this is adding something to the sound that wasn't in the recording, but perhaps this is a better way to reproduce a sound of a specific location/room than simply recording music playing there through a microphone.

 

Here's the IR plot from a vector average of 5 REW sweeps at my sweet spot:

If I play music over speakers you hear the direct sound first, the next several milliseconds of reflected sound is filtered away from our hearing.  This happens at least partly due to directional cues arriving later from a different direction detected by the shape of the outer ear, and from our ears constantly moving slightly (even if we think we are holding them still).  The brain can find reflections that have a similar signature though delayed and ignore those.  

 

If I record the sound at listening position, then the microphone picks up direct sound and reflections.  When I play that back over speakers the delayed and reflected sound comes from the same physical spot.  The effects of our outer ear have no way to separate that from recorded music.  Nor does movement of our head.  So it gets treated as a primary direct sound mixed in with the music and it is not filtered.  Those recorded reflections will create secondary reflections of those in the room, but our ears filter that away.

 

Also as a sidenote REW on IR plots ignores sound after a few milliseconds to clean up the result.  For the same reason our ears do, to filter out reflections in the room and come closer to getting the IR of the speaker as it would manage outdoors or in an anechoic chamber.  You can adjust that time in the settings.  Longer times give better frequency resolution with more chance of reflections being found, while shorter times are cleaner with less frequency resolution. 

[esldude]

10 minutes ago, pkane2001 said:

 

That's true, Dennis. I getting to know pretty well the mechanism for determining direction of the sound. And indeed, our head/ear anatomy is configured to allow some sense of direction to be derived from the incident sound wave. I've been doing a deep dive into this subject since I started using headphones, for about a year now. But I question to what degree the brain/ear system uses directional cues when ignoring reflections (it would make sense that it does, but I've not come across this during my informal research). Do you have any references to studies that demonstrate this effect?

 

And thanks, yes, I'm aware of REW is ignoring later reflections. For my room, it's really not significant as the room is relatively small, and I'm not trying to reproduce every last reflection. After all, I don't hear these anyway when listening through speakers. But it is interesting to me that I can reproduce at least some sense of my speaker/room system through another transducer by applying a measured IR and convolution. Reminds me of optics and image processing using a PSF (point spread function). PSF is very similar to an audio IR function, but in two spatial dimensions.

https://econtent.hogrefe.com/doi/abs/10.1024//1421-0185.58.3.170?journalCode=sjp

 

There are a few ways we hear direction.  Time and intensity differences as well as pinna filtering and head movement.  The brain appears to peg a location for initial direct sound, and as long as the various directional cues agree it blocks out delayed reflections.  As each of the known directional cues are removed, the accuracy of our perception of location is reduced.  I don't know that I can point to research showing directional cues from pinna specifically are used for filtering of reflections, but as pinna directionality is one of the factors in our hearing of location(especially height) it seems it will be involved. 

 

You won't get access to these papers, but here are several abstracts showing there is work in the area regarding the importance of the pinna directional cues

 

https://www.researchgate.net/publication/18781713_The_Role_of_the_Pinna_in_Human_Localization

Sorry, I wish I had something more definitive and in a paper available to read rather than just abstracts. 

 

[esldude]

9 minutes ago, kumakuma said:

 

In this case, you have no way of knowing if the faults that you are hearing are faults in the rig itself or faults introduced by Jana's equipment or recording methodology.

Frank is just practicing reference-less evaluation.

 

Current methods for automatically evaluating sound systems rely on gold-standard references. However, these methods suffer from penalizing sounds that are correct but not in the gold standard. We show that reference-less sound metrics correlate very strongly with human judgments and are competitive with the leading reference-based evaluation metrics. By interpolating both methods, we achieve state-of-the-art correlation with human judgments. Finally, we show that sound metrics are much more reliable when they are calculated at the system level instead of the component level.

 

https://aclweb.org/anthology/D16-1228

 

Some paraphrasing was incurred in the above statement.  

[esldude]

We never did discuss the sound of the video in the original post.  

 

So can we get back on topic.  

 

First, I don't seem to get the real feel with binaural recordings that other people report.  So maybe my reports aren't comparable with those who do.  In this case, it was better than usual in that sound objects seemed on a line from just outside each ear thru the middle of my head.  Often these sound fully in my head and clustered near the top third of my head subjectively.   

 

Next how did it sound?  I preface this with saying I don't think I know what Devore speakers sound like from this.  So while I'm describing what I hear don't conflate it that I'm saying the speakers sound this way.  They may or may not. 

 

The bass was tubby, loose and might not have been excessive had there been a counterbalancing treble.  But the treble was pushed far down to me.  The vocal range and just above (yes I know there were no vocals in places) sounded recessed as if a dip in frequency occurs there.  The upper midrange seemed somewhat excessive though maybe with a fleshed out midrange it would have been mostly okay.  So I'm hearing a tubby uncontrolled bottom end, recessed midrange and annoying upper midrange with no treble to speak of above that. 

 

Over speakers I heard about the same frequency balance with a one dimensional sound strung left to right not unlike the headphones.  The upper midrange bite was not so apparent over speakers.  

 

So using headphones what did you other people hear?

 

PS my headphones were Beyer DT880 Pro and Sony MDR 7510 Pro series.  

[esldude]

11 minutes ago, STC said:

 

Binaural recording is always about hearing what someone heard through their individual pinna. The ear space is usually fixed at 17.5cm and the shape of the head is like an average Caucasian. Furthermore, it also depends your own ears frequency response which can sometimes go very wrong for some people when listening to binaural recordings. Basically, no two men hear alike.

Yes, but some people report amazing life-like 3d hearing from such recordings.  My guess is my pinna are not shaped at all like average or something along those lines.  My guess is good microphones in my own ears might make binaural recordings which are good to me.  Still I wanted to know how the original video linked sounded to others. 

[esldude]

 

I found the binaural recordings of Mitchco better for me.  He made them with small microphones that fit within his own ear.  There was a little more outside my head sound and some sense of space that was lacking in the dummy head binaural.  Over headphones I did find it to have almost a hole in the middle.  Reminded me of listening to jecklin disk recordings I've made over headphones.   It sounded more coherent to me over speakers and also seemed better than the dummy head recordings over speakers.  Again it reminds me of jecklin disk recordings I've done. 

 

There was some sort of variable comb filtering or juddering shimmering effect to the sound which was rather bothersome.  I wonder if that was Mitchco's normal head movement?

 

Maybe @mitchco could join the discussion about that. 

 

For context he may wish to read this post of mine above.

 

and listen to the video made with the dummy head binaural in the initial post in this thread. 

[esldude]

4 minutes ago, STC said:

 @esldudeYou mentioned about head movement and binaural recordings. Care to explain?

My understanding is some part of our ability to dismiss the room is related to small continuous head movements we all make.  I get this from some writings of James Johnston.   And there is also the effect of us turning our head in larger movements.  The thing the Smyth Realizer is said to fix.  

 

I've also read, and wish I kept the paper, where it was said in an anechoic chamber listening to a pair of loudspeakers, the room is dead of course, and you just hear sound from speakers at the speaker mainly.  But if you had the listener place their head on a chin rest and against a forehead rest to arrest small head movements the sound for at least many people suddenly imaged mostly inside their heads similar to what most recordings do over headphones.  

 

So in mitchco's recordings I hear a juddering sound at times.  Most noticeable in the Giorgio track.  It is almost as if each track were on tape and one track or the other is sticking and releasing rapidly.  Or as if one channel is speeding up and slowing rapidly rather like one channel only flutter. So I wonder if this is involuntary head movement by mitchco or some vibration of the mics in his ears or what exactly is it?

[esldude]

So nobody else listened to the binaural files form the video in post #1 of this thread or those from Micthco's article?

[esldude]

1 hour ago, mitchco said:

Hi Dennis,

 

To answer your question, there were 2 issues with the recording. One was the tube amp was switched out with an unknown amp and I heard something similar as you are describing during music playback and of course it got recorded.  Arch heard it too and we did not have any time to troubleshoot the playback issue. The other problem is that the little Sony action cam has a built in automatic gain control that can't be disabled. 

 

Had I done it a 2nd time, I would have brought my own amps and the Hilo ADC would have produced a much better recording. Alas, it was a one time only shot and turned out reasonably good. But you are indeed hearing what you describe. PS. I tried to not move my head during the recording...

 

I need to listen to the other recordings listed in the thread. In the meantime, folks might find these interesting:

 

https://3diosound.com/pages/examples

If the sound was coming from the amp, then it was a form of motor-boating.  I've seen this in tube amps.  Even good designs can sometimes do that if weak tubes are in circuit.  McIntosh 240's if the driver tubes get weak will do something like that. 

[esldude]

1 hour ago, Blake said:

 

Over speakers:  

 

-Bass did seem a bit loose to me as well, particularly on the first track

-Treble was too recessed, to the point of almost being MIA

-No soundstage depth, it was flat and I agree it was similar to a headphone soundstage (although the second track was not as bad as the others)

 

Having said the above, I am totally unfamiliar with the music.  Perhaps as she records other setups, using the same tracks, we can begin to develop some sort of very rough "baseline" subject to multiple caveats.

 

I have no doubt the system sounds different and MUCH better in person. I have not heard DeVore speakers but I know that they are generally highly regarded.

 

I still need to listen on headphones (which is what Jana suggested in her intro).

 

I don't have the expertise on this point, but others in this thread certainly do (esldude, Mitchco, STC)- I am wondering if anyone has any helpful suggestions to provide to Jana that might improve her protocol and perhaps provide them in the commentary to her YouTube Channel?  I really think she should be encouraged and supported in this project.  It may never be perfect, but I think it is a great idea.

 

 

 

Thank you.  

 

I considered responding in her youtube posting.  Also thought it might be good if she took part here.  

 

My problem is I don't want to throw cold water on the project at this early stage. And I can criticize what she has done while not having great suggestions to do better.  At this point I don't know of a good way to remotely record a stereo system so others can listen and get much of an idea of how it sounded.  For speakers the best I can suggest is recording each one fairly close up and you'll get a good representation of its frequency response.  Even then with large speakers like the Devore you may find it difficult to get the drivers to mix in a way similar to hearing it live. 

 

All of which points out a failing of stereo.  In terms of accuracy or the search for the absolute sound, if stereo recording can't reproduce two sound sources of limited dispersion accurately upon playback what hope is there of true accuracy for the far more complex sound of musicians playing in space?  It is all a bit of a tricky contrived illusion.  Which is okay, but not the message many audiophiles want to hear. 

[esldude]

39 minutes ago, Ralf11 said:

maybe the goal to get much of an idea of how it sounded is not possible, but perhaps some correlates of various sound quality factors between 2 different systems could be compared?

There are better ways like measurements of the speakers for that.  You record in two different rooms, with different times of reflection and strength of reflection and you will have a hard time finding out much about it that way.

[esldude]

Maybe a good topic for another thread, or one to continue in this one. 

 

What is the optimum method to record your stereo at your house so others could listen to the recording and know what your stereo sounds like?

[esldude]

1 minute ago, pkane2001 said:

 

As I suggested earlier, record IR at the listening position, then use it to convolve with any piece of music while playing through reasonably flat headphones.

Yes, but is the IR at the listening position going to give someone the correct sound elsewhere?  I know it will give the sound as it was at the listening position, but doesn't that primarily mean we believe FR is the main issue for sound?  I mostly agree with that idea fwiw.  

 

I've not done exactly what you describe, but maybe I should. 

[esldude]

1 hour ago, pkane2001 said:

 

It would be an interesting experiment to try IR's from different systems to hear how they all sound. Anyone who has done measurements of their speaker system at the listening position using REW sine sweeps should be able to create an IR wave file that captures their room and system characteristics. If everyone could then upload their IR file to share with others, we'd have a very interesting collection to play with! All that's required is a player/renderer software that supports convolution (for example HQPlayer) and a set of quality headphones. If there's interest, we could start a separate thread for this discussion.

 

Heck, I might actually finally learn what Frank's sorted system sounds like ?

 

Yes, let us start a thread about that.  I'm interested.  Sounds like fun.  

 

I'm aware of the concept, but have never bothered to actually do such things.  

 

Should we start a thread in the general forum or in DSP, Room Correction???

[esldude]

49 minutes ago, STC said:

 

Why looking for the imperfect IR of listening rooms? Google for free IRs of real concert halls, churches and many others. 

The idea is for me to take the impulse response and let others use it with convolution so they can listen over headphones and get a good idea (we hope) of what listening at my house sounds like.  Ditto for other people. 

 

I've not done this so don't know how well it works.  I know recording at the listening position does not work.  I know recording very close to the speakers does a pretty good job of letting someone hear the FR of my speakers remotely.  Binaural works according to some people though not for me.  So if convolving the IR of my listening position lets someone hear my speakers/room good.  Even better you can choose your own music and know what it might sound like over my system at my house once you have my impulse response to use during playback.  

[esldude]

1 hour ago, STC said:

 

Maybe I am still stuck with the OP intention.

 

What will you be using for convolution? Dirac?

I don't know yet.   I don't have Dirac though a friend does.  So I was hoping for some open source solution.  Maybe the Foobar Impulse Response Convolver plug in?  Maybe a convolver plug in for a DAW?  Have any suggestions?  Oh and we weren't trying to steer away from the OP intentions.  Which is why the talk of a new thread. 

 

Getting back to the OP, if we get this figured out to work better than binaural I see no reason to attempt to get Jana's interest in it. It might be more useful for the same purpose than what she has done in the first video. 

[esldude]

12 minutes ago, STC said:

 

It will never be better than binaural. 

 

BTW, i think there is some confusion about the terminology used here. Maybe, convolution is using real IR to create reverbs. Another way is using artificial reverbs. It is possible, to use IR of the room and inverse the frequency response and use them for room correction. I thought that's what the OP intended but then some of the other replies caused a bit of confusion.

Well to be very clear, binaural is piss poor for me.  If you wished to let people use headphones to hear what the sound was on another system at another location what would you suggest?

 

My complaint was with binaural being so poor.  With a stereo pair at the listening position being so poor.  It was suggested using IR and convolvers would let you hear over headphones a better result to accurately let one hear what another location sounded like. 

[esldude]

7 minutes ago, STC said:

 

Why not burn about $75 and get the Roland binaural microphone and make few recordings. Use the same earphones and listen to the recordings. Theoretically,  you should hear the same ( a little compromise is required). 

 

 

Unless the IR was recorded with a binaural microphones, the room signature will never be the same.

That has been on my list of things to do. 

 

But it doesn't solve the problem here.  Those will hopefully work for me.  But not for someone else.  That is the idea presently under discussion.  How do I record something, so that others can play it back and hear what I heard?

[esldude]

2 hours ago, pkane2001 said:

Would love to hear the process for doing this! Does it involve rewiring?

Well one of Frank's big ideas is soldering all connections.  So that would mean soldering all the connections between neurons.  Which will take some time.  Frank speaks about it taking lots of time.  It also means the normal plasticity of the brain to re-wire its connections would be impacted.  Which might lead to repeating the same thing over and over and over until you didn't realize there was anything else.  

[esldude]

Okay, so getting back to this IR convolution thing, we'd need the IR for left speaker at left ear position, plus a lower level mix of right speaker at left ear position.  This mix would become the left channel?  Then IR for right speaker at right ear position, and a lower level mix of left speaker at right ear position.  Which becomes the right channel.  

 

So does this get us to me hearing your room over my headphones?   Already sounds like there could be so much fine tuning it isn't much better off than binaural.  We would have the convenience of playing any music as we choose thru such processing. 

 

Have any of you done this and have anything to report?  I've listened to 'canned' reverb profiles using convolved IR.  It does give you a sense of another space, but I don't know if that is going to work more or less accurately for remote hearing another system.  

 

Remember my jumping off point was the OP video is not a representation of hearing the Devore speakers.  Nice try, good intentions, but pretty awful.  Or at least I hope so.  If Devore speakers sound that way I don't want any. 

 

 

[esldude]

The best series I've seen in terms of being a bit useful are those done by this outfit SonicSense. 

 

They record 1.5 meters from the speaker with a very flat omni microphone in a studio (presumably with reduced reflections and somewhat dead).  I wish there were more of the music and different types.  I've found with this approach you can hear the basic frequency balance pretty much as you would hear it at your house.  Or more so since your phones will alter the results, it works to hear relative frequency balance between two speakers.  

 

Now there is the idea our brain lets us mostly hear the direct sound of speakers and filter out the room other than low frequency issues below the Schroeder frequency.  So such an approach isn't too bad.  Even if our room had various issues above the Schroeder frequency our brain will ignore it and hear the speaker anyway.  

 

So if these guys instead of recording their music, recorded a sweep to get the IR of the speaker, they could convolve it and give us the result.  After which we could hear it over phones and play the music of our choice.  Presuming our brain would filter out the room whether my room, your room or wherever, it would let us compare speakers usefully over phones.  If the recorded system has any interaction with the speakers like a tube amp altering FR or something that would get included in the recorded IR as well. 

 

This still wouldn't reproduce all info about the speaker in terms of directionality effects.  Nor could it sound like it would in a huge space as in such conditions our hearing in my experience is less able to filter out longer timed reflections, and the sense of a large space.  But comparing basic sound character of a given amp and speaker it might not be missing too much.  The binaural presentation seems useless to me nowhere near what this simple approach can manage. 

 

So thoughts or comments upon this?  If Jenna Dagdagan used this up close approach in future videos would it give us a clearer picture of the sound of someone else's gear?  There is the issue of recording too close on multi-way speakers, and not getting everything the way you hear it at a bit more distance.  

 

I wish the guys at Harman could give us insight into this using their spin-o-rama data.  

[esldude]

3 minutes ago, STC said:

 

It doesn’t filter out although the term “filter” sometimes used in some papers. Indirect sound will always heard and interpreted.  Depends on the arrival time, it gives sense of space and reverbs. As long as the delay is not too long between the preceding sound, it will be interpreted as one. Otherwise, you will hear echoes. 

That sounds like filtering to me.  Yes our ear responds to all of it.  But depending upon arrival time our subjective perception does not hear those early reflections or is only mildly heard.  Which is why I said in very large spaces the various reflected sounds are so late they'll be heard as a sense of space.  

[esldude]

10 minutes ago, pkane2001 said:

 

Don't know how this can be evaluated, unless you compare headphones+IR to your own speaker system. I can share my IR, but how would anyone know if it's at all accurate?

 

I wouldn't fine tune IR with cross-feed, at least not initially, which is what's being suggested here. While it may help a bit with making the sound move out from inside one's head, I find that it tends to muddy the sound. 

 

When I have a chance I'll upload my IR file and you can tell me what you think. No processing was done to it, and I listen to it without cross-feed.

Yes, you can tell us how well it correlates with your sound.  As I can do once I do this on my end.  Then share the IRs.