Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
In Reply to: Re: The reality of figure of 8 microphones: posted by regreene on July 29, 2005 at 14:48:46:
Dear Mr. Greene,
1)“Spaced omni recordings present confusing and internally contradictory locational cues to the listener to the point that things are not exactly anywhere.” Yes you are right, the problem is that as long as we use two speakers, the same applies to our playback system. In other words, any recording done with spaced omnis, ORTF or crossed pairs, played back on two speakers located 10 feet apart, will suffer from these “contradictory locational cues” due to our spaced loudspeaker set up and the fact that we can not prevent people moving around when they listen to music. See also point 3.
2)“the least colored(or should I say coloured) vocal sound was that of the Coles ribbon(a figure 8).” I do not now how this microphone was set up in relation to the voice, but for voice recordings, the distance between microphone and source is rather small, let say around 30 cm. Compare this to a orchestral recording where the microphone distance to the source is often 5 meter or more. When manufacturers give specifications of the frequency response of their directional microphone, they always mention the measurement distance between sound source and microphone. Simply because the output of their microphone, for the low end, can vary up to 20 dB or more when the source to microphone distance varies from extremely close to 5 meters or more.
3)“play a mono pink noise signal through two speakers”, I am happy that you mention this test. This is exactly one of the tests we use for setting up our loudspeaker set up. This mono signal tells you a lot about your loudspeaker set up and the ability of our stereo loudspeakers, spread 8 - 10 feet apart, to produce a non coloured phantom image. To hear this pink noise signal uncoloured you have to listen with one ear closed, and the other ear positioned exactly on the central axis. It will be very difficult to locate the origin of the sound but at least it is uncoloured. If you listen with both ears, you have to sit again exactly on the central axis and now you will hear a nice phantom image in the middle but it is already coloured due to the delayed arrival of the left speaker signal at you right ear, interfering with the non delayed signal from the right speaker, causing a comb filter. Needless to say that the same happens on the other ear. When you move side ways as you suggest, these comb filters will change frequency, and therefore the pink noise will be different coloured depending on your position. Now attenuate the signal on your right channel about 3 dB. As a result the image will shift to the left. Now move again and notice that you will hear the same comb filters as before, less extreme, but they are still there. What we also can hear is a sharp shift of the phantom image position to the left when you move left. This test tells us that our 2 loudspeaker set up is not capable of producing a non coloured, stable phantom image ANYWHERE between the loudspeakers when the signal is a mono signal. As long as we use two speakers for stereo, we will suffer from this unwanted artefact.
Ergo: we have to work with a stereo loudspeaker set up that is far from perfect.
Lets see what happens when we use different main microphone set ups in combination with our spaced loudspeaker set up.
We are listening to a recording made using a coincident microphone set up. The spatial image is completely build upon level differences so the left an right signals are identical (mono) signals. The only thing that varies from left to right is level. Now we have a sound source halfway to the right on stage, which is audible half way to the right on the speaker set up. Now move a little bit off centre to the left. Immediately the phantom image will move to the left. Furthermore, we will have a comb filter as we have seen in our test before. The fact that the comb filter frequency on the playback side is independent of the angle of incident of the sound on the main microphone system is a crucial observation. This comb filter is very strong with deep notches from only one (actually two, since we have two ears,) pronounced comb filter frequency and is the same for direct sound and diffuse sound on the recording. So for an orchestra, spread from left to right between the loudspeakers, the tonal balance of the orchestra and the acoustics of the hall will suffer equally from this single pronounced comb filter when not sitting exactly in the middle since a single comb filter is extremely audible.
Now we look at the spaced microphone set up. Imaging is mainly based on time differences. We start with the same source on stage, halfway to the right, and our phantom image half way to the right on the speaker set up. Sitting exactly in the middle, the direct sound will suffers from a comb filter due to the fact that there is a time difference between left and right. When we move a little bit off centre, the phantom image will also move, but less compared to the coincident set up. The comb filter will change frequency like before. But the comb filter frequency for the direct sound and the diffuse sound will change independent from each other. The reason for this is the fact that at the playback side, the comb filter, which is the result of moving position, is super imposed on the infinite number of diffuse sound comb filters created by the pick up of the spaced omni set up.
The time difference between the left and right signal varies with the angle of incident on the main system and therefore, on the playback side, there is not one single pronounced comb filter but a infinite number of different comb filters. So for the reverb of the hall, the infinite amount of different comb filters will effectively compensate each other and as a result we have a smooth flat frequency response for the diffuse sound field almost independent of the listening position. The comb filter on the playback side is “absorbed” by the infinite comb filters on the recording side. If we look at the orchestra positioned from extreme left to extreme right, the comb filters will also vary, dependant on the angle of incident. For the strings, which are large groups of people, this comb filter will almost not exist. For solo instruments like flute or trumpet, the comb filters will be more pronounced, but at the same time the comb filter notches will be filled by reflections, so at the end, the resulting comb filter is not as extreme as one could expect from theory.
So my conclusion is that although a spaced omni set up is maybe not perfect, it coops much better with the problems we encounter on the playback side. I prefer the less deep and less pronounced comb filters combined with a more stable (not sharp) image compared to the more instable and sensitivity to single frequency comb filters of the coincident techniques. I have tried many different main systems. 15 years ago I have done at least 25 recording with crossed fig of eight and/or ORTF systems before I decided to continue with spaced omni’s.
My personal view is that since we can not prevent listeners moving their head when listening to music, we have to take the shortcomings of our loudspeaker set up in account when discussing microphone set ups.
4)“Omnis pick up diffuse field sound in abundance “, yes, omni’s do pick up a lot of diffuse sound compared to crossed fig. of eight when mounted at the same distance to a sound source. But the crossed fig. of eight are mostly positioned twice (1.7 to be exact) as far away from the sound source compared to omnis, so the direct to indirect (diffuse) sound ratio will be equal for both systems. The last time we did a recording in the Philharmonia in St. Peterburg we had our omni’s 1.5 meter in front and 3.5 meters above the orchestra. During the recording Mr. Alexander did, the main system was 15 feet (5meter) in front of and 10 feet up, more then 3 times as far. So probably there was at least as much diffuse sound ,if not more, on this recording then we had.
I do appreciate your response very much, and I find this a very stimulating discussion. So please, if you find my writing offensive, please remember that I am not English.
I have limited time at this very moment, so I cannot respond to all of Mr. Geijsen's interesting remarks. I shall return to this later, if I may.
But I do want to clarify one point. When I was talking about the moving away from the central axis with the same pink noise signal in each of two speakers, I was trying to give people a way to hear time-delay comb filtering, not commenting on stereo playback itself. This was not very clearly expressed perhaps. The point is that when one listens to the same source with a time delay, as when picking up a live (single) source with spaced out microphones, comb filters are introduced when the microphone signals are combined subsequently.
It is of course the case that stereo playback itself introduces time delays when one moves off the central plane.
Let me suggest a test. Imagine walking across a stage carrying a pink noise source and that this source is going to be recorded.
If you record with a Blumlein pair, there is no time delay and all that changes is the relative intensity of the signal in the two channels. If you record with two (say) spaced omnis, then each microphone will pick up the same signal with slightly changing intensity but very substantial time delays.
Now imagine playback. In the Blumlein, the two speakers are playing each an uncolored signal with no time differential. The ear/brain is receiving no time information but is locatiing the source by intensity stereo. As it happens, up to around 1500 Hz, the signals combine around the head to give almost exactly the phase and frequency response that would have happened in reality. In the higher frequencies, reconstruction is not perfect on account of head shadowing effects, but it is quite good.
Now imagine the spaced omni playback. When the signals arrive at the listening position, there are substantial time shifts. This moves the tonal character around.
You do not have to take my word for this. Buy J. Boyk's(Performance Recordings) test CD. This is a straight forward test. Recordings were made of click sounds from a collection of identical speakers spaced along a line. It is analogous to my walking across the stage.
Blumlein plays back to locate the clicks correctly and ALL THE CLICKS HAVE ESSENTIALLY THE SAME TIMBRE. But the spaced omnis not only locate the clicks in the wrong places, they also makes teh clicks from different locations sound quite different in timbre.
Of course, three mikes(with three channel playback) would do better. And even more would do even better. Indeed according to Huygens' Principle from physics if one had infinitely many mikes and channels and playback speakers one would end up with... Blumlein actually, for all practical purposes.
I know it is complicated to compute what happens with more than two mikes(mixed to two channel playback), although it can be done.
But one really cannot get the comb filters out once they are in.
Moreover, the test is very convincing. Everyone interested in this subject really ought to get this CD and listen.
As to time difference stereo: yes of course time differences shift images. But these time differences in the spaced omni technique are FAR larger than the ones which occur naturally. In fact if one is anything but very far away, they are VERY FAR larger. The differences presented could not occur with a natural sound source ,exceeding as they do the max interaural time-of-difference from a natural source in any location at all(eg that directly to the side).
It is not believable, and in my experience is not true, that such a thing produces anything like convincing and precise imaging.
Of course Blumlein erases the interaural time differences altogether. This might be expected to cause trouble(and it is one argument for ORTF--but not for spaced omnis which are over the top exaggerated in this regard). But surprisingly it seems not to. I have to go now but I discuss this point in my article on Boyk's test CD on my website www.regonaudio.com
To save me typing time I would be most grateful if everyone interested in this discussion would read that article before we continue.
Robert E. Greene
This post is made possible by the generous support of people like you and our sponsors: