Dual mini ear mics for measurements...

[Lanion]

Check out page 47 of this paper (page 47 of the pdf, page 35 of the document). Super cool.

https://www2.ak.tu-berlin.de/~akgroup/ak_pub/abschlussarbeiten/2011/Brinkmann_MagA.pdf

 

[tesseract]

Very cool, thanks for that, Lanion!

 

[Sonnie Parker]

Interesting info Lanion... thanks!

And welcome to AV NIRVANA

 

[aackthpt]

Funny to find this thread, as I started to go down this rabbit hole a few years ago. I think I was thinking of it both as a room measurement method and as a way to make binaural recordings on the cheap. I'd have to see if I can dig deeply enough in the archives to see what sort of microphones I was thinking of using.

Speaking of which, one of the Stereophile CDs back in the day had some Formula1 recorded exactly this way, with ear mics. Yep here it is, it's their test CD 3, track 15. https://www.amazon.com/Stereophile-Test-CD-3/dp/B0000049XR

 

[Matthew J Poes]

Binaural recordings are always made with ear mics as far as I know. That was a cool test track. Sounded disturbing real.

REW can’t be used for two channel measurements because the second channel is only accessible via loop back. For all practical purposes it’s not a 2 channel recorder. I asked John if he planned to add that and he said yes but not soon the only interim fix is to take measurements with Two instances of REW open and then have one in the mode where it looks for an external test tone.

My purpose for dual mic measurements is room characterization. I wouldn’t be shocked if that could be used for dsp processing of the speakers but I am certain DIRAC and other similar corrections cannot make use of that information. It can use multi mic averaged measurements but that’s also what it does with multiple measurement locations. During the beta phase we tested this with a mixer into a usb mic interface.

 

[Matthew J Poes]

http://www.bnoack.com/index.html?http&&&www.bnoack.com/acoustic/bass-ratio.html

Look at IACC for an idea what I wanted to do. I still plan to give this a try. I just need to get accurate measurement mics in a matched pair. Not cheap. I’m seeing of cross-spectrum would match a pair for me. I will not be using USB mics as I want to be sure there is just one clock in the AD converters. The measurement is looking at timing differences in delayed energy so the mics can’t have any timing differences due to clocks not being synced, which they wouldn’t be with usb mics.

This can also be used to set speaker locations more accurately.

 

[Allen Rumbaugh]

I have read and reread your posts here. Am I understanding it correctly that you want to bring in your outer ear into the measurement window? It appears that the mics are mounted facing outward in your ear canal. I just don't see how that is a benefit. I use this measurement mic http://www.earthworksaudio.com/microphones/m-series/m30/ it is hard to fit that quality in your ear. I am used to using Spectrafoo but REW looks like it will do the same thing, that is average multiple pickup points in the room. I would be very curious to see the difference in the measurements between the in ear mics and the m30 at multiple locations.

 

[Matthew J Poes]

I have read and reread your posts here. Am I understanding it correctly that you want to bring in your outer ear into the measurement window? It appears that the mics are mounted facing outward in your ear canal. I just don't see how that is a benefit. I use this measurement mic http://www.earthworksaudio.com/microphones/m-series/m30/ it is hard to fit that quality in your ear. I am used to using Spectrafoo but REW looks like it will do the same thing, that is average multiple pickup points in the room. I would be very curious to see the difference in the measurements between the in ear mics and the m30 at multiple locations.

The M30 is one of the best mid priced measurement mics on the market for sure. I used to have one myself but sold it a few years back.

I’m not personally interested in in-ear mics but it sounds like Sonnie and others are. I have interest in dual mic measurements to characterize the spaciousness/soundstage of the space. It’s more an experiment/curiosity as IACC is a spec for performance spaces and not home theaters.

I think the interest in the dual ear mics is more a belief it may do a better job measuring what we hear. There are a number of researchers doing work like this, as well as a few proof of concept products that used this approach.

REW does average and is in fact the approach I and others use to get the average response of the room over the entire listening area. That would serve a different purpose since timing information and phase are lost in such a measurement.

 

[AudiocRaver]

Playing catch up here.

No news, of course, and it has been referred to already, that the HRTF (hearing response transfer function) is a big factor with in-ear or on-ear or around-the-ear ANYTHING.

Interesting to me is the fact that with regular loudspeakers (or with hearing in general) the HRTF (via the outer and middle ear) is in the chain of what we are hearing all the time, in fact serves a role in the brain's creation of imaging and a soundstage (what direction did that sound come from?),,and with headphones or IEMs it is not, or at least is to a much lesser degree. HRTF is the bridge between speakers and headphones, it compensates a headphone's response to make its measurement have some equivalence with a speaker's measured response, and explains why a flat headphone response (mic at the ear drum) will sound different from a flat speaker response (mic in a free field).

As pointed out, an in-ear mic would have to be really tiny and located in the plane of the ear drum to accurately capture sound like the ear.

 

[Matthew J Poes]

Playing catch up here.

No news, of course, and it has been referred to already, that the HRTF (hearing response transfer function) is a big factor with in-ear or on-ear or around-the-ear ANYTHING.

Interesting to me is the fact that with regular loudspeakers (or with hearing in general) the HRTF (via the outer and middle ear) is in the chain of what we are hearing all the time, in fact serves a role in the brain's creation of imaging and a soundstage (what direction did that sound come from?),,and with headphones or IEMs it is not, or at least is to a much lesser degree. HRTF is the bridge between speakers and headphones, it compensates a headphone's response to make its measurement have some equivalence with a speaker's measured response, and explains why a flat headphone response (mic at the ear drum) will sound different from a flat speaker response (mic in a free field).

As pointed out, an in-ear mic would have to be really tiny and located in the plane of the ear drum to accurately capture sound like the ear.

Well I'm going to revive this thread because I've gotten myself into the rabbit hole that is David Griesinger's body of literature. I took a break from measuring headphones and trying to understand the source of the errors, it was pretty easy to figure that out, and decided to see what else I could use the EARS for. First I used it as the microphone for a video recording of my room as I walked around making sounds to see if it gave a cool effect. As it was Binaural with an HRTF, it should have given a perceived 3D sound and it did. Very cool!

Of course, that wasn't really a scientific study of any kind and there was nothing all that useful there so I decided to instead take 2-channel measurements of a room to see if I could begin to detect LF ITD, per David's work on LF spaciousness and listener envelopment. This has been hugely challenging, far worse than I thought, but... a) I'm making progress in actually measuring a rooms ITD by frequency band, and b) I am successfully making binaural measurements of a room, but I cannot use REW or any of REW's test tones to do this accurately.

What is so cool is that the elusive measurement needed by all of us is a measure of what we hear, not a single point measure of loudness. That is of limited value. Yet no measure like that currently exists, and certainly not one that is implemented in readily available software. This is getting closer. A cheap binaural microphone with an HRTF and a stereo recording of a log-sine sweep with reverb allows you to then create an impulse of the room and assess the rooms effect and interaction with the speakers on creating that sense of SS&I, spaciousness, envelopment, externalization of the source, etc. All things we can't normally measure at all.

I hope to turn all of this into a series of videos and articles, I'm working through it all now. A lot of the other ideas I had for tech articles hit walls and this just seems way more interesting. I think the hardest part is taking these ideas, which are hugely complex, and distill them into a simple and easy to understand set of concept. I think we all care about SS&I, Spaciousness, and envelopment, but I think the average person doesn't understand that they care about it. People don't want mono-sound, they know that, but yet the understanding of source azimuth is totally missing from an average persons vocabulary. That is the gap that we need to bridge.

Wayne if this interests you as much as me, and you want to help, I can certainly use it. I have a bunch of presentations on REW and acoustics coming up over the next 6 months and I hope to bring some of this in so that I can begin to really simplify these ideas and aid in writting the tech articles. I'd like to get the first of the articles out before the end of July.

 

[DanDan]

I reckon it best to keep the intent and practice separate. Binaural Recording is probably best done using a Dummy Head or HATS. IMO it would need to be listened to on Audiology Headphones, pure flat response presented at the ear. Normal headphones are voiced.
Acoustic Measurement practice depends on intent. Do we want to measure what is arriving at the eardrum, or at the mouth of the canal? Or would we be better off Measuring what is happening in the air at the ear location, but without the ear, or body? B&K advice that a measuring mic near a body will experience up to 6dB change in FR around 400Hz. We know that LF response varies wildly at even a 2" distance. So surely it would be wise to measure at two locations about 8" apart, at ear height. We can easily switch speakers on and off to gather L+R, L, R, for both Earlocs. BUT either L+R will not have the head blocking effect. So...... I suggest a Jecklin Disc or something similar. If one is prepared to move the Mic to the Earlocs, all of this can achieved using just one mic.
DD

 

[Matthew J Poes]

I reckon it best to keep the intent and practice separate. Binaural Recording is probably best done using a Dummy Head or HATS. IMO it would need to be listened to on Audiology Headphones, pure flat response presented at the ear. Normal headphones are voiced.
Acoustic Measurement practice depends on intent. Do we want to measure what is arriving at the eardrum, or at the mouth of the canal? Or would we be better off Measuring what is happening in the air at the ear location, but without the ear, or body? B&K advice that a measuring mic near a body will experience up to 6dB change in FR around 400Hz. We know that LF response varies wildly at even a 2" distance. So surely it would be wise to measure at two locations about 8" apart, at ear height. We can easily switch speakers on and off to gather L+R, L, R, for both Earlocs. BUT either L+R will not have the head blocking effect. So...... I suggest a Jecklin Disc or something similar. If one is prepared to move the Mic to the Earlocs, all of this can achieved using just one mic.
DD

DanDan I am referring to binaural impulse response measurements. Not Binaural recordings. It isn’t for listening, it was for studies into small room spaciousness, envelopment, etc. since I have a device that has a Pinna and relatively accurate measurement mics, I see no reason not to use it. Far quicker to do one sweep.

http://www.bu.edu/tech/support/research/visualization/gallery/binaural/
Here is an example discussion of the topic.

Or
https://www.researchgate.net/public..._Using_Frequency-Dependent_Coherence_Matching

I was mostly using it for dereverberization and as an alternative method to get at LEV and Spaciousness that David feels is workable in a small room.

As I noted all of this had to be done outside of REW. In fact, even using the other software I was using couldn’t calculate anything meaningful from this so I had to switch to a Matlab program.

I’ve set this whole experiment aside because once I had all the data I realized that without some point of reference I couldn’t make sense of it. There exists no standards or points of comparisons for rooms with such small dimensions.

The only cool thing I did with it was use it to pinpoint reflections causing specific peaks and dips. So instead of calculating them and guessing which is which, you can tell precisely. You just turn the head around 180 degrees in 5 degree increments and keep sweeping. You can then filter the impulse response around a particular problem and figure out which reflection is responsible and what direction it is coming from. How useful that is I have no idea? I was mostly screwing around at that point to see if it would work.

 

[DanDan]

Hmmm, I agree one sweep is handier, and a blocking element creates a more realistic measurement. But I know of no measurement of envelopment or spaciousness. Relatively accurate mics close to the body and in a Pinna, close to the highly resonant canal, should accurately portray the frequency response at that location. A stereo impulse response should enable one to simulate that response using a convolution engine. I have often stated a plan to take such binaural IRs and indeed Recordings of Ref Trax played over speakers, in stellar Control Rooms. Never gotten around to it, and to hear it properly again would require Audiology or other 'Flat' headphones. I think my point is that by speaker or even by headphone, everything we hear is Pinna and Canal influenced. But if one measures with a particular Pinna/Canal/Blocking effect built in...... what can you use the data for?
DD

 

[Matthew J Poes]

Hmmm, I agree one sweep is handier, and a blocking element creates a more realistic measurement. But I know of no measurement of envelopment or spaciousness. Relatively accurate mics close to the body and in a Pinna, close to the highly resonant canal, should accurately portray the frequency response at that location. A stereo impulse response should enable one to simulate that response using a convolution engine. I have often stated a plan to take such binaural IRs and indeed Recordings of Ref Trax played over speakers, in stellar Control Rooms. Never gotten around to it, and to hear it properly again would require Audiology or other 'Flat' headphones. I think my point is that by speaker or even by headphone, everything we hear is Pinna and Canal influenced. But if one measures with a particular Pinna/Canal/Blocking effect built in...... what can you use the data for?
DD

Well I received a headphone measurement system and I wanted to see what else I could do with it, so this is what I came up with. I really didn't have a big intent beyond trying to think of more interesting ways to characteriize the acoustics of a room and speakers.

I did a bunch of experiments, but in the end the data wasn't all that useful. I don't think it's because the data isn't useful, but because I think you need two things I was missing: A) A strong conceptual model of ideal behavior, and B) points of comparison across other rooms and speakers. I didn't have either so it akin to looking at a picture of a blackhole with no theory to explain what a blackhole is.

As for meausres of LEV and Spaciousness, those are things like IACC and Diffuse Field transfer function.

There is also an image tracking experiment I did that might help identify the accuracy of a speaker/room's ability to reproduce all of these functions since the measurements contain the ITD, ILD, and Reverberation information all in each pair of impulses. It's created by a bunch of paired impulses taken while a tone is panned in some defined increment across the azimuth (say 5 degrees). I did this thinking it might create a way of measuring a speakers imaging, but same problem, no point of reference. Without knowing what it's supposed to look like and correlating it with subjective impressions of imaging, it wasn't so useful. I also feel like the data needs to be graphed in a 3-dimensional graph to make it easier to interpret. I was manually comparing the values for each of these parameters.

 

[DanDan]

I think most CR and HT action is closer to Anechoic than to Diffuse. But I guess the future is most likely in the direction of electronically assisted ambience as well as so called DRC.
DD

 

[DanDan]

Hmmm, got me pondering now, Tx. I hadn't considered those elements of performance spaces, but of course why not.
The CR, with Reflections below -20dB for at least the first 20mS doesn't translate well to Domestic, or really anywhere. There have been decades of struggle trying to make the Control Room Response more 'human' and translation friendly. jim1961 has a long long thread chasing Nirvana, so to speak. Haas Kickers, the lot. He ends up with rear/side speakers and a Lexicon Reverb. Boggy (My Room Acoustics) uniquely does full surround envelopment in Diffusion.
Hmmmm. Happy New Year.
DD

 

[Matthew J Poes]

I think most CR and HT action is closer to Anechoic than to Diffuse. But I guess the future is most likely in the direction of electronically assisted ambience as well as so called DRC.
DD

Yes absolutely and the ability to measure the rooms acoustics more fully including its reverberation and direction of reflections can be critical to optimizing newer forms of DRC in multichannel systems. The two areas I see most emerging out of this is MIMO and wave domain optimization methods.

 

[Matthew J Poes]

Hmmm, got me pondering now, Tx. I hadn't considered those elements of performance spaces, but of course why not.
The CR, with Reflections below -20dB for at least the first 20mS doesn't translate well to Domestic, or really anywhere. There have been decades of struggle trying to make the Control Room Response more 'human' and translation friendly. jim1961 has a long long thread chasing Nirvana, so to speak. Haas Kickers, the lot. He ends up with rear/side speakers and a Lexicon Reverb. Boggy (My Room Acoustics) uniquely does full surround envelopment in Diffusion.
Hmmmm. Happy New Year.
DD

If you are getting at what I think you are getting at, then yeah the research and standards around these concepts are poorly understood in small rooms and it’s possible they already fall into a kind of optimal zone given the very small dimensions.

As noted, I was in contact with David Greisinger and fooling around with a new toy. I don’t attest to have done any great science here. There is something to this but I’m not totally sure what yet. And I’m not sure this is the right avenue either. The more I think about the issues the more I think ambisonic microphones and 3-D impulse measurements are probably what would be needed and a ton more work to correlate this with actual hearing perception to make sense of the results. All above my pay grade at this time.

 

[DanDan]

Yup a New Frontier. Dirac have shown me some amazing results using multiple LF sources with DSP. If the mics are worth it, you very nearly have a Jeklin Disc Rig there. A block of acoustic foam between them... maybe a layer of ply in the middle to ensure blocking and to provide a mic stand mounting. I must check out Dave, I met him yonks ago here. Good chat about his reverbs, I subsequently bought and used a PCM91. But my, and his, all time favourite, is the 480L. That has a magic voice in it.
DD

 

[DanDan]

I am in favour of ear position measurements. A I mentioned, I reckon the optimum way would be a pair of decent omnis and a Jeklin disc. Two channel recordings of each round of L, R, L+R speaker sweeps would deliver measurements including the effects of both speaker signals arriving at both ears, with all the summing and cancellation that entails.
The results can be viewed individually to isolate L, R, differences, listener location specific anomalies. Also, we can average at will in the software. This is an advance over single mic nose measurements IMO.
DD