I thought this might be an interesting discussion.
First, a little background.
Mid-side recording is a stereo microphone setup where there's a main mic pointing straight at the musician(s), and then there's a figure-of-8 (dipole) mic with the null aimed at the musicians.
The idea is that the main mic captures the middle/centre image, and the side (dipole) mic captures the left-to-right stereo difference signal.
By putting that through a patch matrix, you can derive a standard left/right signal, but with a key benefit: you can change the stereo width after the recording has taken place - anything from purely mono to 100% stereo difference signal.
I think it'd be rather cool to have that concept operating in a HiFi system.
I have a few drivers here that I think could work to make a suitable dipole system, although I think it'll need to be a 3-way setup to keep the figure-of-8 pattern nice and symmetrical throughout the operating range.
For kicks, I'll probably do this as a baffle-less design, and I'm planning on a small 3-way setup:
2x Tang Band W6-1139 (100-500Hz) - mostly chosen for mechanical headroom
2x Bose 2.5" full-range drivers (500Hz-3kHz)
2x small dome tweeters. Haven't picked a pair yet, but am looking for 3/4" or 1/2" diaphragms to keep dispersion nice and wide.
By using pairs of drivers facing in opposite directions, they should end up mechanically nice and balanced, so I can put bolts through the mounting holes and hang the next driver below. Distortion cancellation is a bonus, and hopefully it'll even out some of the frequency response anomalies you get when listening to the magnet side of a driver.
That's all for now. Thoughts welcomed - I'll try running the Bose drivers up to 20kHz next week and get some tweeters ordered in due course.
Chris
First, a little background.
Mid-side recording is a stereo microphone setup where there's a main mic pointing straight at the musician(s), and then there's a figure-of-8 (dipole) mic with the null aimed at the musicians.
The idea is that the main mic captures the middle/centre image, and the side (dipole) mic captures the left-to-right stereo difference signal.
By putting that through a patch matrix, you can derive a standard left/right signal, but with a key benefit: you can change the stereo width after the recording has taken place - anything from purely mono to 100% stereo difference signal.
I think it'd be rather cool to have that concept operating in a HiFi system.
I have a few drivers here that I think could work to make a suitable dipole system, although I think it'll need to be a 3-way setup to keep the figure-of-8 pattern nice and symmetrical throughout the operating range.
For kicks, I'll probably do this as a baffle-less design, and I'm planning on a small 3-way setup:
2x Tang Band W6-1139 (100-500Hz) - mostly chosen for mechanical headroom
2x Bose 2.5" full-range drivers (500Hz-3kHz)
2x small dome tweeters. Haven't picked a pair yet, but am looking for 3/4" or 1/2" diaphragms to keep dispersion nice and wide.
By using pairs of drivers facing in opposite directions, they should end up mechanically nice and balanced, so I can put bolts through the mounting holes and hang the next driver below. Distortion cancellation is a bonus, and hopefully it'll even out some of the frequency response anomalies you get when listening to the magnet side of a driver.
That's all for now. Thoughts welcomed - I'll try running the Bose drivers up to 20kHz next week and get some tweeters ordered in due course.
Chris
Did you mean to say two crossfiring dipole mics, I'm having a little trouble picturing how one would decipher which side is which.
You like a challenge
Orbit sound stuff is like this I think: Airsound | Orbitsound
And while you are at it, try this, abit more refined solution with the same idea:
Stereophonic Sound from a Single Loudspeaker
http://elias.altervista.org/html/SingleSpeakerStereo.html
Works well! more speaker drivers required though.
I suggest try with cheap full range drivers first how do you like it before investing too much
And while you are at it, try this, abit more refined solution with the same idea:
Stereophonic Sound from a Single Loudspeaker
http://elias.altervista.org/html/SingleSpeakerStereo.html
Works well! more speaker drivers required though.
I suggest try with cheap full range drivers first how do you like it before investing too much
Last edited:
Chris,
Sounds like you might be after something similar to "E" in this thread:
https://www.diyaudio.com/forums/mul...virtual-single-source-horn-3.html#post4119931
Or the threads mentioned by tmuikku in post #4.
I tried a prototype, the stereo effect was not very convincing, little better than simply processing through a single speaker, so didn't pursue the design further.
Looking forward to hearing your observations.
Cheers,
Art
Last edited:
which pressure gradient microphone brand are you referring to? or are you referring to hybrid back to back cardioids?
Last edited:
Got nothing to do with the brand of mic. Any mic with switchable polar patterns that includes figure of 8 can be used or (almost) any ribbon.
To make the middle&side stereo technique work you record the output of the figure of 8 mic on two tracks and invert one of them as far as I know.
Microphone practice - Wikipedia
To make the middle&side stereo technique work you record the output of the figure of 8 mic on two tracks and invert one of them as far as I know.
Microphone practice - Wikipedia
There appears to be some confusion about the term "mid/side" (often abbreviated to "MS") that describes a recording technique. The notion of an "MS" loudspeaker emerges instead as a variant of the "stereo dipole" that has been used to describe narrow-angle stereo loudspeakers (somewhat unfortunately because all stereo loudspeakers form a laterally-orientated dipole). But the term "MS" speaker invites only more misunderstanding.
Strictly speaking, an "MS" encoding comprises a pressure, omnidirectional (zeroth order) signal (the "M" channel) and a laterally orientated dipolar, velocity (first order) signal (the "S" channel). Conventional "LR" encodings can be interchanged with "MS" encodings without any loss via the simple transformations:
M = sqrt(2).(L+R)
S = sqrt(2).(L-R)
L = sqrt(2).(M+S)
R = sqrt(2).(M-S)
As such, any process acting on one representation can be enacted identically on the other by an appropriately (and linearly) transformed process: The MS and LR representations are just the same information stored differently.
In the recording arena, a variety of "MS" recording techniques can be implemented. An (ideally coincident) omnidirectional and dipolar microphone is an obvious example, but two laterally opposing cardioids or two "crossed" "LR" dipolar microphones can achieve the same result. For (some times) good reasons, engineers often forego such exacting definitions and apply microphone techniques that are not strictly "MS" encodings (but can effect better listening results - we hope!). Of course, studio based recordings often have no such acoustic reference at all: What constitutes high fidelity reproduction is then less easy to define by anyone but the mastering engineer. In short, recording is set to remain an art.
But the reproduction of stereo recordings is more amenable to scientific understanding and different loudspeaker arrangements are no less of a transformation process than any other prior process. But an "MS" speaker as described here opens the pathway to more confusion: Reproducing a stereo recording at a loudspeaker(s) at one point in the listening room is NOT the same as reproducing stereo for a listener situated at another point. In short what you are describing is instead a three-speaker stereo speaker system with a narrow angle between them.
I have to say at this point that my own experiments with such arrangements have been very favourable (and for good reasons). One downside is the need for extra LF boost in the S channel, another is that the listening area where the stereo works well is much smaller than for wider angle set-ups. (Incidentally, the so-called "MS" loudspeaker also happens to be one extreme solution of the three speaker stereo arrangements derived by the late Michael Gerzon to maximise the useful listening area).
So to my point after much needless meandering... It is not an "MS" speaker, it is a narrow-angle three speaker stereo system with inputs S/sqrt(2), M and -S/sqrt(2).
Strictly speaking, an "MS" encoding comprises a pressure, omnidirectional (zeroth order) signal (the "M" channel) and a laterally orientated dipolar, velocity (first order) signal (the "S" channel). Conventional "LR" encodings can be interchanged with "MS" encodings without any loss via the simple transformations:
M = sqrt(2).(L+R)
S = sqrt(2).(L-R)
L = sqrt(2).(M+S)
R = sqrt(2).(M-S)
As such, any process acting on one representation can be enacted identically on the other by an appropriately (and linearly) transformed process: The MS and LR representations are just the same information stored differently.
In the recording arena, a variety of "MS" recording techniques can be implemented. An (ideally coincident) omnidirectional and dipolar microphone is an obvious example, but two laterally opposing cardioids or two "crossed" "LR" dipolar microphones can achieve the same result. For (some times) good reasons, engineers often forego such exacting definitions and apply microphone techniques that are not strictly "MS" encodings (but can effect better listening results - we hope!). Of course, studio based recordings often have no such acoustic reference at all: What constitutes high fidelity reproduction is then less easy to define by anyone but the mastering engineer. In short, recording is set to remain an art.
But the reproduction of stereo recordings is more amenable to scientific understanding and different loudspeaker arrangements are no less of a transformation process than any other prior process. But an "MS" speaker as described here opens the pathway to more confusion: Reproducing a stereo recording at a loudspeaker(s) at one point in the listening room is NOT the same as reproducing stereo for a listener situated at another point. In short what you are describing is instead a three-speaker stereo speaker system with a narrow angle between them.
I have to say at this point that my own experiments with such arrangements have been very favourable (and for good reasons). One downside is the need for extra LF boost in the S channel, another is that the listening area where the stereo works well is much smaller than for wider angle set-ups. (Incidentally, the so-called "MS" loudspeaker also happens to be one extreme solution of the three speaker stereo arrangements derived by the late Michael Gerzon to maximise the useful listening area).
So to my point after much needless meandering... It is not an "MS" speaker, it is a narrow-angle three speaker stereo system with inputs S/sqrt(2), M and -S/sqrt(2).
Last edited:
Then it is very good way of constructing a stereo reproduction system for which I have yet to hear anything better - even if it is only practical for a single listener.
But it does require attention to the prior processing stages that are required...
Firstly, the S-channel LF boost can be significant (and additional to that added to compensate for microphone spacing or even just to add 'spaciousness').
Secondly, if the dipole spacing is close enough, cross-cancellation can be avoided althogether and simple HF boost used to compensate the inherent fall-off prior to the first (then non-audible) "coincidence dip". Further to this, if you are EQ'ing in the digital domain, you can add the requisite S-channel delay for precise compensation.
Of special note here is that, unlike more normal two loudspeaker "LR" cross-cancellation schemes, the M-channel (and centre loudspeaker) needs no such processing. Even for wide-angle stereo, this can offer a significant performance gain.
You only record the figure 8 mic to one track, ( why waste another track recording the same thing) but you split this track on replay. Any mixer will work as a MS matrix. Send the figure 8 to 2 strips, hit the phase button on one of them, and then pan them left right till you get the stereo spread you want and adjust the mid mic level for more or less direct sound.
Last edited:
Yep, and I'm planning on deriving the "side" signal from conventional stereo, and playing that signal through a dipole speaker. The method of making stereo from a mid/side system is reversible.
In the case of my own recordings, I have the benefit of avoiding messing with the patch matrix stuff and just feeding mid/side as the mics picked up in the first place.
While the EQ required to get a dipole system to be flat full-range can be incredible, I suspect there won't be much low-frequency output required of the dipole array - most music is mono (or very close) below 100Hz, so the main "mid" speaker will handle that.
Stay tuned - should have results in the next few days.
Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.