Mid-side stereo techniques

[soundbloke]

I think I got confused thinking it could actually work!

Please don't misinterpret my objections to the theory though, the narrow-angle three speaker arrangement with MS inputs can perform outstandingly (IMHO better than any other arrangement I have encountered for a single user). Just understanding what you are building helps when it comes to make it work well

 

[scottjoplin]

Thanks for clearing things up for me, much obliged.

 

[Pano]

OK, are we talking about 3 speaker stereo with a matrix, or are we talking about 2 speaker with a dipole speaker on edge and a cardioid speaker facing front?

 

[soundbloke]

OK, are we talking about 3 speaker stereo with a matrix, or are we talking about 2 speaker with a dipole speaker on edge and a cardioid speaker facing front?

A dipole is two acoustic sources - hence the three "speaker" stereo.

And there is no requirement for a front facing cardioid pattern, which if implemented would mean the array was not an "MS" array, but would also render it not information preserving - or "valid" to quote a term used earlier in this thread. But then we are moving into the realm of second-order directivity functions and so many more opportunities for confusion...!

 

[6vheater]

Think of it as multiple disparate random noise sources rendered coherent by ONE brain with 2 spheres, making up for the errors from 2 not quite perfectly similar ears, 2 not quite perfectly similar ear cavities with resonant freqs around 2khz, unequal sensitivities and phase responses caused by outer ear formations, with a lot of bone and tissue (pinna) which also transfers lots of vibration (the 3rd ear).

Think of concorde coming roaring towards you, then passing overhead and behind you with giant doppler effects and crackling from ultrasonic/supersonic wavefronts.

You can place it all dead accurate spatially with 2 ears...
then imagine how much more difficult it would be if you had 3 brain spheres and 3 ears to cope with...

Dr Tomatis made a living from people who were confused, and had relational issues.
Hearing psychology and treatment were a central element in getting Depardieu on his feet, or he would not be famous today.

The relationship between hearing, acuity and behavioural psychology are strongly linked.
Hence why buildings nowadays are starting to be designed with sound and ambience as a primary driver for good productive working evironments....
Not many people know that.

Much as I can see how theory "could" return a result from a dipole (the reverse of a Decca triangle), you could never get a concept of depth and distance,- only width (which is the achilles heel of course of Decca TR).
Which achilles heel do you want to experiment with?

I'm sure Avery Fisher tried all this stuff 50-60yrs ago.
He was the one getting musicians to play behind a screen, as well as a hifi replay system.
The audience had to blind guess which was the real one...

 

[soundbloke]

Much as I can see how theory "could" return a result from a dipole (the reverse of a Decca triangle), you could never get a concept of depth and distance,- only width (which is the achilles heel of course of Decca TR).
Which achilles heel do you want to experiment with?

Any stereo loudspeaker forms a dipole in respect of S-channel information. It is the differences between the arrival times from the (nominally) two loudspeakers forming the dipole that imparts the phase information at the ears relevant to the stereo effect (strictly an LF effect).

A narrow angle set-up requires boosting the S-channel to restore the signals at the ears that a wide angle would produce. What is the advantage of such a complication? Simply that the boosted narrow-angle option can reduce the comb filtering apparent in the wide-angle version. For narrow enough angles, the first comb null can be sufficiently high in frequency to allow complete compensation (another boost but possibly with the ideal delay compensation too). And if you add a centre M-channel, then combing in the often important vocals panned dead centre doesn't exist!

The Achilles' Heel of any spaced microphone array is incoherence - just that many people prefer it as it adds an effect similar to that apparent in a reverberant field, but at the expense of smearing the temporal information for which coincident techniques excel. But the narrow-angle three-speaker dipole is still best placed to resolve a stereo recording no matter how well it has been encoded. With proper S-channel compensation and even extra boost, coincident recordings can produce depth, width, distance and so on at the maximum possible fidelity from stereo. And not just in theory

 

[6vheater]

Any stereo loudspeaker forms a dipole in respect of S-channel information. It is the differences between the arrival times from the (nominally) two loudspeakers forming the dipole that imparts the phase information at the ears relevant to the stereo effect (strictly an LF effect).

Except surround systems...

I don't use stereo, I use enhanced stereo with 4 channels (2 rear ones).
It's night and day compared with boring 2 speaker systems.

 

[soundbloke]

Except surround systems...

I don't use stereo, I use enhanced stereo with 4 channels (2 rear ones).
It's night and day compared with boring 2 speaker systems.

That is because you haven't tried the arrangement yet With the S-channel processing sorted, you will be amazed at what narrow-angle, three speaker stereo reproduction is capable of.

 

[6vheater]

err.. where to start?
We actually author surround sound recordings.
When done right they're absolutely fantastic, but infamously difficult to reproduce correctly.

I am one of the few people who actually has a genuine Matrix H decoder from the BBC and a big pile of recordings that used it. (the system much criticised by Gerzon in fact for the experimental prom season broadcast in surround in 1976).

Being as I have just spent the whole day vetting other sound engineers work on the monster system I have and designed, I doubt I would consider changing it.

You can listen to two of our recordings on one of my posts on the previous pages.
You might like it.

We are currently working evaluating some new Russian microphones.
It's rather busy right now.
There's a constant to and fro of exchanges and comments with the manufacturers v what other people have claimed to discover, then listening to what they have done or not done as may be.....
EXHAUSTING!


Oh and here are Gerzon's calrecs...gotta love the local chem lab clamps!Another threesome!
You can read a lot more about him hereHave fun!

 

[Pano]

A dipole is two acoustic sources - hence the three "speaker" stereo.

Well that's confusing. If I have an open baffle speaker with cones, or an ESL or Magnepan I don't consider it two speakers. It's a single speaker that radiates out both sides. Similar to using a Beyer M-130 to record the S signal. I would not consider that 2 mics - just one mic that picks up from both sides in a figure 8. I understand what you mean by 2 sources, but it's confusing when talking about speakers (or mics).

And there is no requirement for a front facing cardioid pattern,

How so? What would you use? Figure 8? Omni?

 

[soundbloke]

Well that's confusing. If I have an open baffle speaker with cones, or an ESL or Magnepan I don't consider it two speakers. It's a single speaker that radiates out both sides. Similar to using a Beyer M-130 to record the S signal. I would not consider that 2 mics - just one mic that picks up from both sides in a figure 8. I understand what you mean by 2 sources, but it's confusing when talking about speakers (or mics).


How so? What would you use? Figure 8? Omni?

A dipole is defined as two identical, spaced sources operating out-of-phase. An open baffle microphone or loudspeaker driver is an approximation to that.

M is defined to be zeroth-order, omnidirectional (same as W in B-format), although often is not in microphone arrays.

 

[soundbloke]

err.. where to start?
I am one of the few people who actually has a genuine Matrix H decoder from the BBC and a big pile of recordings that used it. (the system much criticised by Gerzon in fact for the experimental prom season broadcast in surround in 1976).

My museum piece is an original NRDC prototype decoder - it would be interesting to see whose is the oldest

But I shall have a listen tomorrow for sure...

 

[Pano]

although often is not in microphone arrays.

Indeed it is usually a cardioid or hyper-cardioid mic so that the M signals come from one direction.

 

[Robbintip]

Well that's confusing. ,,,

It's a single speaker that radiates out both sides.

The sound wave can be sent in one direction, or the other, by the transient attack of the membrane. But never to two opposing directions by the two opposing poles at the same time.

Similar to using a Beyer M-130 to record the S signal. I would not consider that 2 mics - just one mic that picks up from both sides in a figure 8. I understand what you mean by 2 sources, but it's confusing when talking about speakers (or mics).

Try to think of sound as directional waves.
A dipole mic can receive attacks from two directions.
A dipole speaker can attack in two directions.
A kickdrum can... no it can not compare.

Looking at how a kickdrum operates it's easy to understand it is not mandatory that a musical wave is as beautifully mirrored and organised like a sine wave.

I found it confusing at first too, I was trying to devide infinity (figure of 8) by 0 (dipole) lol. I imagined that there is good use for some problems because a singular solution for all problems must be the problem too.

 

[6vheater]

My museum piece is an original NRDC prototype decoder - it would be interesting to see whose is the oldest
But I shall have a listen tomorrow for sure...

I scarcely use mine in years.
I am waiting to replace a DD vinyl turntable to restart it all.

FYI:-
I constructed the decoder entirely myself from plans from Wireless World, bought the boards from a company that made them specially and bought all the chips from Toshiba (I think) at the time.
It was quite expensive to do.

All components were 1%, and you can change/optimise the channel balance, which I did by placing headphones across the 2 RR/RF channels, then LR/LF.

All recordings from Nimbus were recorded in this UHJ format.
I believe they were used the Calrec SF mics, to make this.

I bought loads of their recordings.
They were very proud of their concert hall, piano and recording techniques at the time inc their S.A.M (a 45 RPM LP), and direct to disk, no editing.

Many years later I listened to all this again, and realised the recordings are actually pretty lousy, but some of the artists OK to excellent, the Steinway well sorted, esp the Beethoven sonate cycle with Roberts.

I much prefer the (EMI) Rostropovich Tchaikovski set which was done in SQ, at Kingsway hall, in 1977, -sadly demolished as a derelict wreck in 1998.

The problem with meddling too much with a well sorted reproduction set-up,-
it's easy to screw it up, taking some time to optimise, especially as I am using complex arrays with bi/tri wiring,with the OPT as active/passive crossover network.

There's a small number of people going to all this trouble, it appears.
The world's a small place.

Objectivity is a highly prized goal.
I am constantly wondering what it is people are after in studios.

 

[soundbloke]

Nimbus put the engineer first to such an extent that they built their own concert hall so they could tell everyone else to sit! I have a few of their discs dotted around too, although I seem to remember being one Haydn disc short And if building your own concert hall was not enough, the horn they built for transcribing 78s was barmy! My memory is of a Soundfield mic at the end of it, but I am sure I remember them using a discrete arrangement - certainly in their earlier recordings. My memory fades... Although I am sure when all else has decayed in the world, there will still be an enormous horn left standing somewhere in the Wye Valley

 

[Pano]

The sound wave can be sent in one direction, or the other, by the transient attack of the membrane. But never to two opposing directions by the two opposing poles at the same time.

Sure. It's always going to compress on one side, rarefy on the other. Then you have the monopole M speaker in the center which will be doing something different.

 

[6vheater]

Nimbus put the engineer first to such an extent that they built their own concert hall

Fascinating stuff.
Thanks for a trip down memory lane.

I have listened to a lot of their stuff, and TBQH pretty much underwhelmed.
I have never been there, but as you do, drove past it countless times.
From what I could understand it became some sort of weird "snob value"...but at a time when eg. Hugues Cuenod was absolutely unknown in the UK.

Wales does have some world famous studios, but mostly for loud rock.
One of them a farm, where originally they had to move the cattle first...

You are reminding me yet again, I have to transfer my collection of vinyl to 24-96 to my local LAN server.... grief! More work!

I do keep saying, look up Avery Fisher's work on reproduction.
He did have the advantage of building some of the best valve amplifiers ever made, as well as inviting musicians to play at his house.

 

[chris661]

Well, I got it all set up again for round two.

Laptop source into M/S encoder into the amplifier, and then out to the speakers.
This time, it's the 8" 2-way as the "mid", and a pair of 4" 2-way boxes as the "sides".

I've played with spacing and levels, and I'm afraid I just can't get anything useful here.


Here's what I'm hearing as I turn up the "side" level with music playing:

- Sets off as mono, of course
- Still sounds mono-ish with the "sides" at "subtle" levels (muting the "mid" confirms levels). Woodbury's description of a "halo of sound around the center" is accurate.
- Go too far, and I hear the stereo FX and panned instruments coming out of a spaced dipole that I'm sat off-axis to. Lots of my own room.

I tried playing pink noise panned hard L or hard R, and could never get it to sound like it was coming from where it was supposed to. It was always centre + some wall splash if you increase the "side" levels.


Once again, I'm convinced that the key to this sort of reproduction is coincident drivers. However, I don't have any interest in pursuing it further. It would've been nice to have a single cube that would create a nice wide stereo image, but that just isn't happening.

Chris

PS - Firing the "side" speakers 45-degrees forwards helped a little, but putting the main speakers back to stereo is a sigh of relief.

 

[soundbloke]

Removing the obvious compromises in your set-up would have been a good start. How did you compensate for the directivity of the dipole drivers? How did you implement the "levels"? I would guess not via the single pole filters that are necessary? Why did you not try the HF compensation and delay correction? Your sigh of relief returning to wide angle stereo is a sign that your method failed to remove the stereo combing. What you do with your own time is your own business, but I would suggested that you conclusions are somewhat premature.