This was posted in a public forum by Richard Lee. I am simply copying it here without edits: Intro I'd like to state my somewhat heretical view of this. I designed the Mk4 Soundfield Mike more than 20 years ago and was R&D manager for Wharfedale Loudspeakers for a long time. I have designed Ambisonic decoders and encoders and Digital EQ for loudspeakers; also a lot of stuff on measuring loudspeakers and applying the new fangled digits to this. My specialty is integrating loudspeakers with electronics and to me, a surround sound / Ambisonic system is simply a loudspeaker. My obsession is listening tests and how to apply these as a measuring tool for design. the Microphone Problem Let's imagine a musical event happening in a hall with good acoustics. A lucky or rich punter with a Mk1 Human Head has the pick of the seats and finds a spot which to him sounds best. Sound from each singer, player or instrument proceeds in an infinite number of different directions, with different timbre characteristics in each direction. Some of these are reflected and modified by the walls and objects in the hall, some many times and may or may not end up at the Mk1 HH. The effect of these is to lead him to say "This sounds great". Note that in a great hall, only a small (less than 10%) part of what the Mk1 HH will experience comes directly from the singers, players or instruments, even if our punter likes a front row seat. If you want to hear how important this is, arrange a musical event in your local anechoic chamber where you will hear only the direct sounds. How can we record the aural sensations he experiences so we can reproduce it at some later date? We take away the Mk1 HH and replace it with an instrument, which we shall call a microphone. What signals should it record to capture our musical event as heard from our favourite spot? Several sets of signals are possible but each of these can be derived from the other sets. The 4 WXYZ outputs of a Soundfield mike are one convenient and complete set. How we use this set is another issue, the Loudspeaker Problem, and may be the subject of a different post. eg Michael makes the point that 4 'microphone' signals (3 for horizontal) is necessary and sufficient and more would MAKE LOCALISATION WORSE. If you have tried the 'music in an anechoic chamber' experiment, you will understand why it is important to get 'flat ratio of free-field to diffuse-field response' but what you really want is FLAT FREQUENCY RESPONSE IN ALL DIRECTIONS. The Soundfield is the closest to this ideal. Because I can define the performance of a perfect microphone, I can say the Soundfield is the best microphone in the world as I know exactly how far from perfection it is. Even very expensive microphones are poor off-axis and using them as a stereo pair makes the situation much worse. But how does the 'perfection' of a Soundfield manifest itself? Stereo Recording I have watched many experienced recording engineers use a Soundfield for the first time and some of you may remember doing this. If he is recording something like a piano, he will have his favourite arrangement, fairly close to it. If he is recording a group in a good hall, he may have a preference for crossed cardiods or hypercardiods at some distance or height. Little by little, he will move the microphone back and little by little, he will dial in more fig-of-8. Eventually, he ends up with the Soundfield at where he would sit himself and on fig-of-8s at 90 degrees! At this point, he usually says, "Nobody uses this arrangement and its definitely too far away" and goes back to the start. The idea that you might want to place your microphone where it sounds best to you is novel because other microphones don't sound best there. Does that tell you something about the Soundfield? Don't you WANT to favour ambience when you use a Soundfield in say the Wigmore Hall? Even if eventually you use a much close position? I think it was Peter Craven who said that, in theory, only a perfect omni or fig-of-8s at 90 degrees would capture the sound of a good hall correctly. Any other arrangement colours the sound because they are biased towards certain directions. Richard Elen (Studio Sound) and John Atkinson (HFN&RR) both made similar surprised statements when they used the Soundfield for the first time. Then when you do move closer, you find the localisation with a Soundfield.is very different from other arrangements. Localisation with a Soundfield is not pin-point for the simple reason, real life is not pin-point. (for pin-point and a flat collage, use panpot stereo) Even a small instrument like a violin has a definite size. Richard Elen (or perhaps Angus McKenzie) said that it was the first microphone he'd used that captured the real life 'pools of sound' which is the audible signature of a string quartet in a good hall. If you listen to Mike Skeet's recording of a drum kit on the first HFN&RR test CD, you will hear how big sound sources are big, and small sound sources are small on a Soundfield recording ... as they are in real life. Low Frequencies The Soundfield has the flattest LF of any velocity mike. The closest is the old STC 4038 ribbon (still a very good mike). That's the reason why they are so sensitive to vibration, wind, doors banging etc. Overload is always after the matrixing. The mike and capsules do not overload until a very high level. (If W is not distorted, the chain up to the matrix has not overloaded) This is one of the difficulties in velocity (including cardiod) capsule design. You need a 'slack' diaphragm for good LF response (simplified explanation and there are other factors) but then the diaphragm collapses and / or overloads easily. On the Soundfield we cheat by using EQ but the capsule design must be good in the first place. One of Michael's criteria for good capsules. LF are a very important part of good ambient sounds and practically all present velocity mikes are rolling off by 100 Hz. I shall leave noise and the absence or otherwise of electrolytics etc. out of this discussion though I obviously have strong (biased) views. However, the above is quite objective. Richard Lee |
Las Vegas Pro Audio 7320 Smoke Ranch Road Suite G Las Vegas, NV 89128 Phone: 702.307.2700 fax: 702.314.2764 sales@lasvegasproaudio.com    |