Hi all,
I've liked binaural recordings for a while. Not the weird YouTube videos, but ones where you've got musicians playing around you and it sounds like you're right there (links at the bottom).
Well, one of the bands I work with would be perfect for that sort of thing, and they're up for it.
Shopping list:
2x Behringer ECM8000
1x polystyrene head
2x latex ears (the ones I picked up were for acupuncture practice).
I had one of the mics lying around, and picked up a 2nd for cheap on the usual auction site.
Sliced the head in half horizontally, cut channels for the mics etc, gaffer taped it all together and had a listen.
The result was, without doubt, an auditory disaster. No bass, horrendous peak in the mid-high range.
As usual, I did some measurements. First up, with my reference headphones, Shure SRH840s. To me, they sound very clean and neutral. Cans by Sennheiser, Sony, Audio-Technica of similar price point all sounded wrong to me.
Anyway, I'm not here to go on about headphones.
Here's a graph.
So that's put the headphones on the head, sweep through the headphones and run the head-mic into REW.
I tried EQing that response flat, which sounded half-decent, but not great. Then I did some reading around with terms like "Diffuse Field" among others.
Next up, I put the good measurement mic (Beyerdynamic MM1) up in front of a Bose MusicMonitor, generated a calibration file, and then connected up the head.
Of course, the frequency response is angle-dependent, so here's a few graphs.
I took an average of those three, which looks like this:
Then I EQ'd that to be pretty much flat, and had a listen.
Much much better. Not perfect, but certainly closer than I got by measuring with headphones.
My phone doesn't want to talk to my laptop at the moment, but when I get that sorted I'll upload a couple of photos of Ed. He looks like a polystyrene head that's been stabbed with a couple of measurement mics, and that's pretty much what's happened.
One day he might get a wig.
That's all for now. I might try playing guitar while walking around and upload a recording.
I also thought it might be a good way of evaluating speakers, but I don't think its quite "there" yet, so that'll be a while off.
Chris
PS - recordings links:
https://www.youtube.com/watch?v=ecOrBqQAuXg
https://youtu.be/itLxXeyM2aM?t=58s
I've liked binaural recordings for a while. Not the weird YouTube videos, but ones where you've got musicians playing around you and it sounds like you're right there (links at the bottom).
Well, one of the bands I work with would be perfect for that sort of thing, and they're up for it.
Shopping list:
2x Behringer ECM8000
1x polystyrene head
2x latex ears (the ones I picked up were for acupuncture practice).
I had one of the mics lying around, and picked up a 2nd for cheap on the usual auction site.
Sliced the head in half horizontally, cut channels for the mics etc, gaffer taped it all together and had a listen.
The result was, without doubt, an auditory disaster. No bass, horrendous peak in the mid-high range.
As usual, I did some measurements. First up, with my reference headphones, Shure SRH840s. To me, they sound very clean and neutral. Cans by Sennheiser, Sony, Audio-Technica of similar price point all sounded wrong to me.
Anyway, I'm not here to go on about headphones.
Here's a graph.
An externally hosted image should be here but it was not working when we last tested it.
So that's put the headphones on the head, sweep through the headphones and run the head-mic into REW.
I tried EQing that response flat, which sounded half-decent, but not great. Then I did some reading around with terms like "Diffuse Field" among others.
Next up, I put the good measurement mic (Beyerdynamic MM1) up in front of a Bose MusicMonitor, generated a calibration file, and then connected up the head.
Of course, the frequency response is angle-dependent, so here's a few graphs.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I took an average of those three, which looks like this:
An externally hosted image should be here but it was not working when we last tested it.
Then I EQ'd that to be pretty much flat, and had a listen.
Much much better. Not perfect, but certainly closer than I got by measuring with headphones.
My phone doesn't want to talk to my laptop at the moment, but when I get that sorted I'll upload a couple of photos of Ed. He looks like a polystyrene head that's been stabbed with a couple of measurement mics, and that's pretty much what's happened.
One day he might get a wig.
That's all for now. I might try playing guitar while walking around and upload a recording.
I also thought it might be a good way of evaluating speakers, but I don't think its quite "there" yet, so that'll be a while off.
Chris
PS - recordings links:
https://www.youtube.com/watch?v=ecOrBqQAuXg
https://youtu.be/itLxXeyM2aM?t=58s
I have been looking at binaural recording and pondered if you record from the ear canal then you should replay at the ear canal i.e. .......using ear buds not headphones . I have yet to try it myself to prove the theory.
That's a good point. IIRC the two ECM mics matched pretty closely, within one or two dB throughout the range. Something to check, though.
Good point regarding earbuds vs over-ear headphones. I do find the effect pretty convincing with those headphones on other recordings, though. I've been meaning to get hold of some good earbuds for a while, so I might have another look around.
Chrs
Good point.
Our pinnae (outer ears) have an equalizing function which headphone designers (usually) take into account. For this reason good headphones never have a flat measured response.
https://en.wikipedia.org/wiki/Auricle_(anatomy)
However with dummyhead recordings that equalsation is already recorded and in this case one would really need flat measuring 'phones.
I tried this, too, using a styropor hat-shop head and electret capsules. The head has no ears to start with, but i modelled flat rims hinting at ear mussles and fixed the capsules to their centers, not working the head itself. I stood this thing at my bed side where i used to sit, pressed record, went out of the door, closed it and came in again. Replaced the artificial head by mine and wound the tape back to when i had left the room. Put good open headphones on and pressed start. I could have sworn, someone just opened the door. 😀
Did yo listen to your recordings thru headphones, speakers or both? Was the tonal balance off?
I used some in-ear mics often back in the 80s and got nice results, even on speakers.
You can also try a simple barrier of disc between the two mics as a test. That might let you know better what's going on. Don't expect flat, it won't be.
I used some in-ear mics often back in the 80s and got nice results, even on speakers.
You can also try a simple barrier of disc between the two mics as a test. That might let you know better what's going on. Don't expect flat, it won't be.
Chris, very cool!
A couple of resources in case you have not seen before:
While these are expensive https://3diosound.com/ Jeff sells the ears separately: https://3diosound.com/products/silicone-ears I have a set and they work very well.
I have had good sound with these: http://soundprofessionals.com/cgi-bin/gold/item/MS-TFB-2
And just in case: http://www.davidgriesinger.com/poster.jpg Also attached is his binaural techniques.
I have been contemplating using binaural mics for digital room correction, but still have a few too many projects on my plate.
Cheers, Mitch
A couple of resources in case you have not seen before:
While these are expensive https://3diosound.com/ Jeff sells the ears separately: https://3diosound.com/products/silicone-ears I have a set and they work very well.
I have had good sound with these: http://soundprofessionals.com/cgi-bin/gold/item/MS-TFB-2
And just in case: http://www.davidgriesinger.com/poster.jpg Also attached is his binaural techniques.
I have been contemplating using binaural mics for digital room correction, but still have a few too many projects on my plate.
Cheers, Mitch
Hi Mitch,
Thanks for the links.
From what I can tell, the frequency response needs to be EQ'd flat for the diffuse field situation. That is, when sound is arriving from all directions simultaneously. Attached is a PDF from Neumann when they were developing one of their earlier dummy heads.
The short story is that if you EQ so that the entire pickup area is flat, there is no directional preference (as there would be if one EQ'd the response flat for a source directly in front of the head).
Now, I don't have an echo chamber to do this work, so I plan to do it manually:
- Set up a small full-range loudspeaker at a distance of around 3' from the head.
- Recording signal from one ear at a time, run a frequency sweep through the speaker.
- Rotate by 22.5 degrees, repeat sweep.
- Once each ear has been through 360 degrees of rotation, change the height of the speaker, and repeat. Probably a few heights, but I don't want a ridiculous amount of data.
- When all the data has been collected, mathematically average the results.
- Apply EQ for each ear seperately. This should account for differences between the two mics etc.
Not a 5-minute process, but if I get something good out of this then I'll be very pleased.
Chris
Thanks for the links.
From what I can tell, the frequency response needs to be EQ'd flat for the diffuse field situation. That is, when sound is arriving from all directions simultaneously. Attached is a PDF from Neumann when they were developing one of their earlier dummy heads.
The short story is that if you EQ so that the entire pickup area is flat, there is no directional preference (as there would be if one EQ'd the response flat for a source directly in front of the head).
Now, I don't have an echo chamber to do this work, so I plan to do it manually:
- Set up a small full-range loudspeaker at a distance of around 3' from the head.
- Recording signal from one ear at a time, run a frequency sweep through the speaker.
- Rotate by 22.5 degrees, repeat sweep.
- Once each ear has been through 360 degrees of rotation, change the height of the speaker, and repeat. Probably a few heights, but I don't want a ridiculous amount of data.
- When all the data has been collected, mathematically average the results.
- Apply EQ for each ear seperately. This should account for differences between the two mics etc.
Not a 5-minute process, but if I get something good out of this then I'll be very pleased.
Chris
Hi Chris,
Very cool. Thanks for the paper. I eq'd the response flat from a source directly in front of the head and got better results than no eq. But I think you are going down the right path to eq the entire pickup area. I will be very interested to see/hear your results!
Mitch
Very cool. Thanks for the paper. I eq'd the response flat from a source directly in front of the head and got better results than no eq. But I think you are going down the right path to eq the entire pickup area. I will be very interested to see/hear your results!
Mitch
So, the testing is done, for now. I only did one height, with the speaker level with the ear. The sweep/rotate bit didn't take long, but I'm impatient.
Here's the average responses for each ear:
Darker blue is the right ear.
Next up, I played with the EQ in REW and got it pretty flat with four filters:
And soon I'm gonna have a listen.
Chris
Here's the average responses for each ear:
An externally hosted image should be here but it was not working when we last tested it.
Darker blue is the right ear.
Next up, I played with the EQ in REW and got it pretty flat with four filters:
An externally hosted image should be here but it was not working when we last tested it.
And soon I'm gonna have a listen.
Chris
Still not great.
Okay-ish through the midrange, but the entire balance is out - everything sounds gutless like a speaker with no BSC. Bringing in the lower midrange helps a bit, but it's not quite right.
With ~15dB boost towards the lower midrange and bass, I'm starting to find the noise levels inherent to the mics a problem, too.
Chris
Okay-ish through the midrange, but the entire balance is out - everything sounds gutless like a speaker with no BSC. Bringing in the lower midrange helps a bit, but it's not quite right.
With ~15dB boost towards the lower midrange and bass, I'm starting to find the noise levels inherent to the mics a problem, too.
Chris
I am not so sure I would trust the measurement that you made with the head enough to try to EQ your headphones with them. Bass may be going right through the head.
This site's measurement looks completely different.
https://www.headphone.com/products/shure-srh840
If I were you, I'd try to measure like the demonstration here, with a board and an artificial ear canal.
John Conover: Using the Panasonic WM61A as a Measurement Microphone
search for 'figure xxxi' on the page and look at the picture just above it 😉
This site's measurement looks completely different.
https://www.headphone.com/products/shure-srh840
If I were you, I'd try to measure like the demonstration here, with a board and an artificial ear canal.
John Conover: Using the Panasonic WM61A as a Measurement Microphone
search for 'figure xxxi' on the page and look at the picture just above it 😉
one of the flaws with attempts at binaural recordings is the assumption that one must reproduce a human head complete with ears.
the human ear and a microphone are not direct equals. microphones do not have a Fletcher/Munson response curve or compensation built in!
i acheived my best results with a three inch spacing and a 160 degree angle, no head or barrier.
recording each as a mono track, panning and level on play back has a profound effect in tailoring the mid response mimicing the F/M curve.
the human ear and a microphone are not direct equals. microphones do not have a Fletcher/Munson response curve or compensation built in!
i acheived my best results with a three inch spacing and a 160 degree angle, no head or barrier.
recording each as a mono track, panning and level on play back has a profound effect in tailoring the mid response mimicing the F/M curve.
I played with condenser mics in my ears and playback through regular phones. My experience was similar to your door. My friend came in, talked to me. On playback the realism was startling.
Sent from my Hi8 using Tapatalk
Sent from my Hi8 using Tapatalk
The best binaural recordings I've heard are with dummy heads, so that's the avenue I'd like to explore.
As a little update, I tried dialling in those EQs and still didn't like it. Put something similar in on the graphic EQ and got something really quite good. Strange, but I'm not gonna complain. I'll be taking it around to a musician-friend's house who's interested. Might do a bit of recording.
Chris
As a little update, I tried dialling in those EQs and still didn't like it. Put something similar in on the graphic EQ and got something really quite good. Strange, but I'm not gonna complain. I'll be taking it around to a musician-friend's house who's interested. Might do a bit of recording.
Chris
That's interesting. Can you expand on that?
The other day I was testing a good old AT2020 mic with Fostex headphones on. Just listening to my wife talk in the kitchen. Once the level was set, the difference between miked and direct was tiny. The mono mic did collapse the room sound a bit, but OK for a single voice. I was surprised at how close they were.
So your stereo mix is intriguing. Please tell us more.
not much more to it other than assigning it to a stereo subgroup. any difference in phase between the two produces a cancellation/reduction. the distance and angle are the important factors.
same thing as frequency nulls between closely spaced tom mics without gates.
same thing as frequency nulls between closely spaced tom mics without gates.
Pano
here's a link to an article that uses a related principle
Q. Why do I sometimes see singers using two microphones on stage? |
here's a link to an article that uses a related principle
Q. Why do I sometimes see singers using two microphones on stage? |