Has anything like this ever been attempted?
To clarify, the test would probably look like this:
The blind listener is in a room with both a live source (someone speaking/singing or playing instruments) and some loudspeakers that play recorded material (recorded from the same source). The listener has to identify whether he's listening to the "live sound" or the playback through speakers.
I'm guessing the biggest challenge for transparent speaker playback would be related to directionality and room effects. (No need to be limited to a stereo setup, though.)
Not quite the same, but:
http://www.aes.org/e-lib/browse.cfm?elib=11885 (http://www.aes.org/e-lib/browse.cfm?elib=11885)
(http://i216.photobucket.com/albums/cc73/AJinFLA/Absolute%20Listening%20Test.jpg)
Has anything like this ever been attempted?
To clarify, the test would probably look like this:
The blind listener is in a room with both a live source (someone speaking/singing or playing instruments) and some loudspeakers that play recorded material (recorded from the same source). The listener has to identify whether he's listening to the "live sound" or the playback through speakers.
I'm guessing the biggest challenge for transparent speaker playback would be related to directionality and room effects. (No need to be limited to a stereo setup, though.)
The biggest challenge is getting the live music to be a close enough duplicate of the recording. I can pretty well guarantee that careful listeners will notice minor changes and timing and intonation. If the differences are small but still reliably audible, the listener perceives that there is a difference, but is unsure what the difference is. In the real world no performer does exactly the same thing twice.
The diagram that AJ found in the annals of the AES (good job, AJ!) was IME nothing new when it was published in 1982, People had been talking about this sort of a test for years if not decades before. It is good example of how to do it right(er) by using the identical same performance.
The remaining challenge is finding matching anechoic chambers and microphones that are good enough. Anybody with a lot of spare space and a willingness to write a 7-figure check can play the anechoic chamber game, but I don't know about the microphones. I've never seen or heard a microphone that would pass the ABX frequency response matching criteria.
This all serves to show how many light years the performance of electronics, especially good digital is advanced over even the best things acoustical.
Not quite the same, but:
http://www.aes.org/e-lib/browse.cfm?elib=11885 (http://www.aes.org/e-lib/browse.cfm?elib=11885)
(http://i216.photobucket.com/albums/cc73/AJinFLA/Absolute%20Listening%20Test.jpg)
Interesting. Do some loudspeakers (and mics) manage to pass the test, aka being indistinguishable?
I can imagine they would, even with that extra mic. There's a mic for the speech anyway (so there's some "coloring" by default) and the listening position is more fixed compared to my live example where the listener could move his head and influence the sound input/direction.
The biggest challenge is getting the live music to be a close enough duplicate of the recording...
Sure, but I didn't have exact duplicates in mind. You'd have various individual performances/recordings and the listener would make a guess every time after hearing one ("live or recorded?").
Acoustic Research used to give regular "live vs. recorded" demonstrations all the time back in the 1960's. They used a string quartet recorded in an open field. They also used AR speakers (equalized, though). The quartet would play for awhile and then would pretend to play while the speakers took over. As I recall it used to fool a lot of people, though it was hardly double blind.
Wharfedale also gave some live vs recorded concerts using an entire orchestra back in the 1950's.
Actually I think it goes back even further if memory serves, Edison gave live vs recorded events using his original gramophone and the listeners of the day could not tell the difference between the gramophone and the live singer. It may have been someone else than Edison though, my memory comes from a book I read many many years ago. Here's a link that seems to confirm my memory but I have no idea if the article has any real authority behind it. http://www.audiocircle.com/index.php?topic=82842.0
Hearsay, or possibly even didn't-hearsay, but I think I remember reading of a such a test in phonograph or gramophone days where it was said that the musicians were coached to sound like the recording.
Interesting. Do some loudspeakers (and mics) manage to pass the test, aka being indistinguishable?
Yes. The problems with your type of proposed test, is the recording system won't capture what your 2 ears hear in the same room, or close to what your ears hear, unless it's something like this: http://www.aes.org/e-lib/browse.cfm?elib=9136 (http://www.aes.org/e-lib/browse.cfm?elib=9136). So right off the bat, you would be asking the speaker(s) to reproduce what isn't there, for sake of comparison.
cheers,
AJ
http://seanolive.blogspot.nl/2010/07/why-live-versus-recorded-listening.html
After a 1916 live-versus-recorded demonstration in Carnegie Hall, the New York Evening Mail stated “the ear could not tell when it was listening to the phonograph alone, and when to actual voice and reproduction together.
This might be a clue (together)
Anyways it proves that audio journalism in 1916 was already at the same high standards we enjoy today :)
Such demonstrations are documented in this book
'Perfecting Sound Forever"
https://www.amazon.com/dp/B0030CHEYU/ref=dp-kindle-redirect?_encoding=UTF8&btkr=1
Interesting. Do some loudspeakers (and mics) manage to pass the test, aka being indistinguishable?
Probably not. To be acoustically transparent (indistinguishable from perfect) a device must have better frequency response and lower distortion than even the best speakers and microphones.
The detailed specification of what frequency response, noise and distortion must be for sonic transparency is complex, but it is usually shorthanded as being frequency response +/- 0.1 dB 20 Hz - 20 KHz, less than ca. 0.1 % THD or IM and noise 80 dB down or better.
Doable with analog electronics and digital, but very challenging to say the least, for anything that operates in the acoustical domain.
The biggest challenge is getting the live music to be a close enough duplicate of the recording...
Sure, but I didn't have exact duplicates in mind. You'd have various individual performances/recordings and the listener would make a guess every time after hearing one ("live or recorded?").
I doubt what you could learn very much in that context. Every time you hear a difference, how do you prove that its not due to your sloppy setup? I say that having actually tried it. My days aren't long enough for that sort of inherently flawed screwing around.
I do recall a Gearslutz conversation in which people were waxing very lyrical about the speakers being really like having the musicians in the room. This might have been studiophiles, but it was professionals, not audiophiles, and earth was regained with a bump when someone asked if anyone in the conversation could ever seriously not know, blind, if they were listening to musicians or their monitors.
But hyperbole can be fun.
You could probably fool me in a concert hall or music hall where the room has a big effect on the sound. However, I have been to plays where the music was "canned" (usually with live vocals) and I've usually been disappointed. You could certainly fool me at a stadium rock concert where the virtually all of the live sound comes through the P.A.
In my living room I think it would be a lot more difficult to foo me. I wouldn't want a rock band in my living room and a full-orchestra wouldn't fit... I wouldn't want the sound of a full-orchestra either. ;)
In a couple of months I intend to do a single blind live vs. recorded session at a hifi show. An internet search brought me to this topic. I've been an occasional lurker for years and I decided it was time to join. I really like that Hydrogenaudio is very science-based.
For the test I'm partnering with an audiophile record label and I have access to an anechoic chamber. We will be using very accurate and quite powerful active monitor speakers. I believe we are well-equipped to make it work!
Of course, this is a marketing stunt. Our aim is to demonstrate the expertise of the recording professionals as well as the quality of the audio system. Therefore, we hope to achieve that the audience will not hear whether they are listening to live musicians or playback from an anechoic recording. But we don't want to use any tricks, such as having the musicians emulate their recorded selves in any way.
Mr. Krueger, let me start by saying thanks for your ABX comparator and I wish your website were still online. I should have saved the audio files on it!
Above you said you had some experience with this sort of inherently flawed screwing around. I understand that live vs. recorded demo's have their limitations, yet I'd still love to make it work. Would you please share your experience and perhaps give some tips?
I would highly appreciate all comments.
yet I'd still love to make it work. Would you please share your experience and perhaps give some tips?
What kind of music, and is there any sound reinforcement? ...Obviously, it will be easier to fool the listeners if the "live" sound comes from a PA system.
It would help to have some distance between the performers and the audience. I'm not sure how much space you'll have at a hi-fi show unless you have access to a theater. And to make it "proper", you should be in a theater, or music hall, or whatever is appropriate for the type of music being performed.
And of course, do your own experiments before going to the trade show.
Has anything like this ever been attempted?
To clarify, the test would probably look like this:
The blind listener is in a room with both a live source (someone speaking/singing or playing instruments) and some loudspeakers that play recorded material (recorded from the same source). The listener has to identify whether he's listening to the "live sound" or the playback through speakers.
I'm guessing the biggest challenge for transparent speaker playback would be related to directionality and room effects. (No need to be limited to a stereo setup, though.)
The biggest challenge is getting the live music to be a close enough duplicate of the recording. I can pretty well guarantee that careful listeners will notice minor changes and timing and intonation. If the differences are small but still reliably audible, the listener perceives that there is a difference, but is unsure what the difference is. In the real world no performer does exactly the same thing twice.
The diagram that AJ found in the annals of the AES (good job, AJ!) was IME nothing new when it was published in 1982, People had been talking about this sort of a test for years if not decades before. It is good example of how to do it right(er) by using the identical same performance.
The remaining challenge is finding matching anechoic chambers and microphones that are good enough. Anybody with a lot of spare space and a willingness to write a 7-figure check can play the anechoic chamber game, but I don't know about the microphones. I've never seen or heard a microphone that would pass the ABX frequency response matching criteria.
This all serves to show how many light years the performance of electronics, especially good digital is advanced over even the best things acoustical.
You don't need to worry about performance differences, since the listener could be played a single long piece and switch between it at points. As the piece continued the listener would be able to switch source back and fourth between live and recorded. If the piece was never repeated then you'd be unable to examin what one was recorded.
And just for the record, anechoic chamber hire is around £1500 per day in the U.K. It's not cheap, but it's not breaking the bank in the name of experimentation.
The microphone would kill you tho, I would assume none are transparent like that. I do have numerous microphones with a flat frequency response, but they tend to sound "bad". I can't quite understand why, but I would guess they are very directional
Interesting. Do some loudspeakers (and mics) manage to pass the test, aka being indistinguishable?
Probably not. To be acoustically transparent (indistinguishable from perfect) a device must have better frequency response and lower distortion than even the best speakers and microphones.
The detailed specification of what frequency response, noise and distortion must be for sonic transparency is complex, but it is usually shorthanded as being frequency response +/- 0.1 dB 20 Hz - 20 KHz, less than ca. 0.1 % THD or IM and noise 80 dB down or better.
Doable with analog electronics and digital, but very challenging to say the least, for anything that operates in the acoustical domain.
Hold on, didn't you previously state that this type of transparency wasn't achievable on analogue equipment? And also, +/- 0.1dB seems
extremely tight for a human to decipher any audible difference. I'd like to see a source for those numbers.
In a couple of months I intend to do a single blind live vs. recorded session at a hifi show. An internet search brought me to this topic. I've been an occasional lurker for years and I decided it was time to join. I really like that Hydrogenaudio is very science-based.
For the test I'm partnering with an audiophile record label and I have access to an anechoic chamber. We will be using very accurate and quite powerful active monitor speakers. I believe we are well-equipped to make it work!
Of course, this is a marketing stunt. Our aim is to demonstrate the expertise of the recording professionals as well as the quality of the audio system. Therefore, we hope to achieve that the audience will not hear whether they are listening to live musicians or playback from an anechoic recording. But we don't want to use any tricks, such as having the musicians emulate their recorded selves in any way.
Mr. Krueger, let me start by saying thanks for your ABX comparator and I wish your website were still online. I should have saved the audio files on it!
Above you said you had some experience with this sort of inherently flawed screwing around. I understand that live vs. recorded demo's have their limitations, yet I'd still love to make it work. Would you please share your experience and perhaps give some tips?
I would highly appreciate all comments.
I'm told that this was done way back then with shellac 78s and horn gramaphones, and achieved amazing results. I'm also told that the "live" performers had special training to sound like the record. Yes: a marketing stunt.
Even the phenomenon of stereo itself requires a certain
willing suspension of disbelief.
I recalled accounts of Acoustic Research's live vs recorded demos, and while I was hunting for more info, I found that Sean Olive had already chimed in on the matter:
http://seanolive.blogspot.com/2010/07/why-live-versus-recorded-listening.html (http://seanolive.blogspot.com/2010/07/why-live-versus-recorded-listening.html)
Interesting. Do some loudspeakers (and mics) manage to pass the test, aka being indistinguishable?
Probably not. To be acoustically transparent (indistinguishable from perfect) a device must have better frequency response and lower distortion than even the best speakers and microphones.
The detailed specification of what frequency response, noise and distortion must be for sonic transparency is complex, but it is usually shorthanded as being frequency response +/- 0.1 dB 20 Hz - 20 KHz, less than ca. 0.1 % THD or IM and noise 80 dB down or better.
Doable with analog electronics and digital, but very challenging to say the least, for anything that operates in the acoustical domain.
Hold on, didn't you previously state that this type of transparency wasn't achievable on analogue equipment?
Please Read carefully for a change. You should distinguish between
analog electronics which is what I said and
analog media which appears to be what you are thinking about. To clarify,
analog electronics includes things like modern op-amp based SS electronics (amplifiers, crossovers, etc.) that outperforms the specs I mentioned, often by orders of magnitude.
Analog media includes LPs and analog tape, which most definitely do not perform to the standards I mentioned.
And also, +/- 0.1dB seems extremely tight for a human to decipher any audible difference. I'd like to see a source for those numbers.
Since you have repeatedly refused to honor such requests from me, back at you. :-(
For the polite and sincere lurkers around here, the best information I have on this topic is in this AES paper:
Clark, David L., "High-Resolution Subjective Testing Using a Double-Blind Comparator", Journal of the Audio Engineering Society, Vol. 30 No. 5, May 1982, pp. 330-338.
http://www.aes.org/e-lib/browse.cfm?elib=3839
You don't need to worry about performance differences, since the listener could be played a single long piece and switch between it at points. As the piece continued the listener would be able to switch source back and fourth between live and recorded. If the piece was never repeated then you'd be unable to examin what one was recorded.
Obviously, you've never done any serious ABXing for subtle differences. :-(
The music between any two different points in a performance is usually
different music, and is generally pretty easy (on the ABX scale of extreme listener sensitivity to differences) to distinguish from the music between any other two different points.
People who don't have real world experience with hearing subtle differences that are still audible would likely never know this. It is one of those things like hearing +/- 0.1 dB level or frequency response differences that you have to personally do, to believe.
Mr. Krueger, let me start by saying thanks for your ABX comparator and I wish your website were still online. I should have saved the audio files on it!
You can still find them on the Wayback machine:
https://archive.org/web/
There were 3 PCABX.COM web sites that were linked together under the covers to take advantage of promotional deals on web sites of a certain size in that day: www.pcabx.com, www.cdabx.com and www.pcabx-pr.com.
You don't need to worry about performance differences, since the listener could be played a single long piece and switch between it at points. As the piece continued the listener would be able to switch source back and fourth between live and recorded. If the piece was never repeated then you'd be unable to examin what one was recorded.
Obviously, you've never done any serious ABXing for subtle differences. :-(
The music between any two different points in a performance is usually different music, and is generally pretty easy (on the ABX scale of extreme listener sensitivity to differences) to distinguish from the music between any other two different points.
People who don't have real world experience with hearing subtle differences that are still audible would likely never know this. It is one of those things like hearing +/- 0.1 dB level or frequency response differences that you have to personally do, to believe.
Believe it or not, many professional musicians can keep tempo. Even with the use of a metronome if need be.
Unfortunately I don't have access to that paper. Can you please post an excerpt? In my personal experience, people can only begin hearing differences around +/-1dB. My ego is not tied to having super elite hearing.
Secondly I didn't provide measurements because I was telling you that I found it to be transparent - what good would measurements be?
Lastly, do you need to be so verbose that you quote me twice? :-)
we hope to achieve that the audience will not hear whether they are listening to live musicians or playback from an anechoic recording. But we don't want to use any tricks, such as having the musicians emulate their recorded selves in any way.
How many and what type of musicians? How many encode/reproduction channels?
You don't need to worry about performance differences, since the listener could be played a single long piece and switch between it at points. As the piece continued the listener would be able to switch source back and fourth between live and recorded. If the piece was never repeated then you'd be unable to examin what one was recorded.
Obviously, you've never done any serious ABXing for subtle differences. :-(
The music between any two different points in a performance is usually different music, and is generally pretty easy (on the ABX scale of extreme listener sensitivity to differences) to distinguish from the music between any other two different points.
People who don't have real world experience with hearing subtle differences that are still audible would likely never know this. It is one of those things like hearing +/- 0.1 dB level or frequency response differences that you have to personally do, to believe.
Believe it or not, many professional musicians can keep tempo.
You apparently vastly underestimate how tight ABX comparisons actually are.
Common musical tempos range from 20 to 140 BPM. The fastest of these is less than 3 beats per second.
With ABX, it is pretty common to reliably detect timing errors in music in the range of 1-10 milliseconds. The coarsest of these is 6,000 BPM.
Get back to me when you can find a metronome and musician that can work consistently that precisely.
Please post relevant recordings. We'll get back to you with ABX logs.
You don't need to worry about performance differences, since the listener could be played a single long piece and switch between it at points. As the piece continued the listener would be able to switch source back and fourth between live and recorded. If the piece was never repeated then you'd be unable to examin what one was recorded.
Obviously, you've never done any serious ABXing for subtle differences. :-(
The music between any two different points in a performance is usually different music, and is generally pretty easy (on the ABX scale of extreme listener sensitivity to differences) to distinguish from the music between any other two different points.
People who don't have real world experience with hearing subtle differences that are still audible would likely never know this. It is one of those things like hearing +/- 0.1 dB level or frequency response differences that you have to personally do, to believe.
Believe it or not, many professional musicians can keep tempo. Even with the use of a metronome if need be.
Unfortunately I don't have access to that paper.
Sure you do. They are available to the public and many corporate and university libraries have them. But if you bought one of your own, it might cost you a fraction of what analog media costs today. :-(
One rule of audiophilia appears to be: "Spend the big bucks on High End audio, but not a penny for common audio test gear or science-based audio information"
Can you please post an excerpt?
(https://hydrogenaud.io/index.php?action=dlattach;topic=107570.0;attach=8249;image)
In my personal experience, people can only begin hearing differences around +/-1dB.
A natural consequence of lack of experience with serious scientific audio testing.
My ego is not tied to having super elite hearing.
My ego is tied to making true, relevant statements.
Secondly I didn't provide measurements because I was telling you that I found it to be transparent - what good would measurements be?
They might convince you of what you apparently don't want to believe.
Mr. Krueger, let me start by saying thanks for your ABX comparator and I wish your website were still online. I should have saved the audio files on it!
You can still find them on the Wayback machine:
https://archive.org/web/
There were 3 PCABX.COM web sites that were linked together under the covers to take advantage of promotional deals on web sites of a certain size in that day: www.pcabx.com, www.cdabx.com and www.pcabx-pr.com
Thanks!
yet I'd still love to make it work. Would you please share your experience and perhaps give some tips?
What kind of music, and is there any sound reinforcement? ...Obviously, it will be easier to fool the listeners if the "live" sound comes from a PA system.
It would help to have some distance between the performers and the audience. I'm not sure how much space you'll have at a hi-fi show unless you have access to a theater. And to make it "proper", you should be in a theater, or music hall, or whatever is appropriate for the type of music being performed.
And of course, do your own experiments before going to the trade show.
The plan is to do cello, acoustic guitar and a small percussive set such as bongo's or djembe drums. Besides that we'll do spoken voice and perhaps singing. We'll probably also do some 'effects' with with a tambourine, or keys.
The room in which we'll be doing the test/show is 8.20 x 6.20 meters. Because listeners will be seated over the entire width of the room, we'll use a minimum of panning between speakers. In order to get a credible width in the soundstage, we'll be using a left, center and right setup.
The recordings will be done in an anechoic chamber. The speakers we'll be using are uni-directional (only to the front), but still radiate relatively wide. These speakers have no front-wall reflection but the musicians in the room do. Therefore we'll use a reflector panel behind the musicians while recording in the anechoic chamber. We'll make photo's and a video ;) .
I intend to do part of the demo 'blind', so I'll need an acoustically more or less transparent screen or curtain. Any tips as to what I could use?
Thanks!
Martijn
I intend to do part of the demo 'blind', so I'll need an acoustically more or less transparent screen or curtain. Any tips as to what I could use?
There is a standard theatrical tool called a scrim that is used for this purpose. You can control its apparent optical transparency with lighting. If you want people to see the scene behind it, you light the scene brightly. If you want it to disappear, you drop the lighting of the scene, and increase the lighting where the spectators are.
Hoi Martijn,
it seems that you have your test ready. The anechoic recording seems the way to go. Please let us know the results.
https://youtu.be/-LfPzgfPXyM (https://youtu.be/-LfPzgfPXyM)
The anechoic recording seems the way to go.
Using how many mics, of what types, placed where?
3ch stereo?
None the less, should be interesting, though not sure what it will show ;)
The anechoic recording seems the way to go.
Using how many mics, of what types, placed where?
3ch stereo?
None the less, should be interesting, though not sure what it will show ;)
While I don't have anything like an anechoic chamber at my disposal, I've done recordings in quasi-anechoic environments. One such environment involved mattresses wrapped around the recording space (which was small) and one is simply out of doors and a goodly distance from any reflective surfaces. Soft earth and to a lesser degree foliage are relatively non-reflective.
BTW AJ if you are interested, I recently posted a comment on what I think may be one of your comments on the AES web site. It is currently in the approval process.
Using measurement mics that were designed to be flat responding in an acoustically environment of "Random Incidence" the resulting recordings tend towards being hot, dry, and bright - similar to very close micing.
Several recordings of natural sounds that were on the old www.pcabx.com web site (still available via the Wayback machine) were made this way.
Attached: one such recording
Using how many mics, of what types, placed where? 3ch stereo?
It looks like they have put a speaker in the anechoic room as well, which could mean that they recorded the cello live and, with the same recording setup, the reproduced sound from the speaker. I did see a Merging Hapi AD/DA converter for a split second, which probably means that live and reproduced audio should be sample sync. Nice for an ABX.
Using how many mics, of what types, placed where? 3ch stereo?
It looks like they have put a speaker in the anechoic room as well, which could mean that they recorded the cello live and, with the same recording setup, the reproduced sound from the speaker. I did see a Merging Hapi AD/DA converter for a split second, which probably means that live and reproduced audio should be sample sync. Nice for an ABX.
So then it would be a mixed live/pre-recorded performance. Now we should be concerned with the provenance of the recorded track.
So then it would be a mixed live/pre-recorded performance. Now we should be concerned with the provenance of the recorded track.
We'll have to wait for more info from Martijn, but I expect this is what they did:
1) record a cello (live) in an anechoic room (file A)
2) replace the cello by the speaker and play back file A. Record the speaker's output with the same mic/pre/room as 1. (file B)
3) Make sure file A and B are properly synched and level matched, and perform an ABX test.
File B will contain twice the effect of the room (probably negligible in this case) and mic/pre/ADC, plus once the speaker's noise+distortion.
In order to get a credible width in the soundstage, we'll be using a left, center and right setup.
Matrixed or discrete?
Nice speakers Btw
Using how many mics, of what types, placed where? 3ch stereo?
It looks like they have put a speaker in the anechoic room as well, which could mean that they recorded the cello live and, with the same recording setup, the reproduced sound from the speaker. I did see a Merging Hapi AD/DA converter for a split second, which probably means that live and reproduced audio should be sample sync. Nice for an ABX.
Yes, I watched video after work, it does appear they may have done so.
I am curious about the 3ch format. For PSR, thats a good start. But the mics and arrangement being used to capture the soundfield....
I recently posted a comment on what I think may be one of your comments on the AES web site.
I haven't posted in a while. What paper?
So then it would be a mixed live/pre-recorded performance. Now we should be concerned with the provenance of the recorded track.
We'll have to wait for more info from Martijn, but I expect this is what they did:
1) record a cello (live) in an anechoic room (file A)
2) replace the cello by the speaker and play back file A. Record the speaker's output with the same mic/pre/room as 1. (file B)
3) Make sure file A and B are properly synched and level matched, and perform an ABX test.
File B will contain twice the effect of the room (probably negligible in this case) and mic/pre/ADC, plus once the speaker's noise+distortion.
Why not just measure anechoic impulse response of the loudspeaker so you can convolute it with any recording (anechoic or not) and compare original vs convoluted (check sync and level as usual) ?
I recently posted a comment on what I think may be one of your comments on the AES web site.
I haven't posted in a while. What paper?
The hd meta study.
Not approved yet.
I intend to do part of the demo 'blind', so I'll need an acoustically more or less transparent screen or curtain. Any tips as to what I could use?
There is a standard theatrical tool called a scrim that is used for this purpose. You can control its apparent optical transparency with lighting. If you want people to see the scene behind it, you light the scene brightly. If you want it to disappear, you drop the lighting of the scene, and increase the lighting where the spectators are.
Thanks Arnold! I should have checked in sooner. I've been trying to find the right material and couldn't really find anything. This seems to be perfect!
Hoi Martijn,
it seems that you have your test ready. The anechoic recording seems the way to go. Please let us know the results.
https://youtu.be/-LfPzgfPXyM (https://youtu.be/-LfPzgfPXyM)
Hi Kees,
Thanks for sharing the video. Will you come and join us for the demo? It is at the XFI Premium hifi show: http://x-fi-audioshow.nl
So then it would be a mixed live/pre-recorded performance. Now we should be concerned with the provenance of the recorded track.
We'll have to wait for more info from Martijn, but I expect this is what they did:
1) record a cello (live) in an anechoic room (file A)
2) replace the cello by the speaker and play back file A. Record the speaker's output with the same mic/pre/room as 1. (file B)
3) Make sure file A and B are properly synched and level matched, and perform an ABX test.
File B will contain twice the effect of the room (probably negligible in this case) and mic/pre/ADC, plus once the speaker's noise+distortion.
Hm... that's actually a great idea! Unfortunately we didn't do that, though. The 8c's were brought to the anechoic chamber for monitoring purposes only. We did a recording of a cello, spoken word an snaredrum. The plan is to compare those recordings one-on-one with the live sound. Exciting!
Using how many mics, of what types, placed where? 3ch stereo?
It looks like they have put a speaker in the anechoic room as well, which could mean that they recorded the cello live and, with the same recording setup, the reproduced sound from the speaker. I did see a Merging Hapi AD/DA converter for a split second, which probably means that live and reproduced audio should be sample sync. Nice for an ABX.
Yes, I watched video after work, it does appear they may have done so.
I am curious about the 3ch format. For PSR, thats a good start. But the mics and arrangement being used to capture the soundfield....
The recordings were done with two microphones. However, we'll mostly work with mono from a single speaker. The playback system will be LCR.
From what I've heard so far, voice and snare drum are very live like. The cello sounds impressive, but it's immediately clear that it's not live. I think the problem is the complex radiation pattern of the cello. Therefore you can't capture the true character of the instrument with a single mic. Play that imperfect recording back over accurate speakers with constant directivity, and it's just not the same thing. We'll be using the recording anyway and it's up to us to try and explain the above to the listeners.
I think it will be a fun demo!