When we think of stereo recording, left/right stereo typically comes to mind: two independent channels of audio, one carrying information from the left of the soundstage, the other carrying information from the right. There is another way to record in stereo, however — one that gives you plenty of options during post-production.
FIG. 1: The middle-side stereo technique, also referred to as mid-side or M-S, employs a directional (mid) mic aimed at the source and a bidirectional (side) mic positioned to pick up sound coming from the sides. The signals must be properly decoded to create the final stereo image.
The mid-side (M-S) stereo recording technique puts the center of the soundstage (mid) in one channel and the side information in the other. These mid and side channels can be adjusted and reconfigured to accurately represent a conventional left and right stereo image (on their own, they do not give you a left/right image). The M-S technique gives you more control over the width of the stereo spread than other miking techniques do, and you can make adjustments at any time after the recording is finished.
Surprisingly, the M-S technique is one of the oldest concepts in stereo audio. The theoretical basis for it appears in Alan Blumlein's seminal 1934 patent, although it wasn't until the stereo experiments of Danish State Radio engineer Holger Lauridsen in the 1950s that the technology caught up with the theory.
Although M-S is most commonly used as a microphone technique, it can also be used on stereo recordings in a mastering situation. It has long been used in broadcasting, largely because properly recorded M-S is always mono-compatible. M-S continues to be a popular technique for studio and concert recording, and its convenience and flexibility make it a good choice for field recording.
Although it may seem complex at first, the M-S miking technique is surprisingly logical and low-tech. The basic concept is that one microphone acts as the center, or middle, channel, picking up sounds from the front of the soundstage. Because the mid mic is aimed straight ahead and panned center, equal amounts of it are sent to the left and right sides of your mixer's main bus. Typically, a cardioid mic is chosen for the mid mic, but any mic can be used.
The side microphone must be bidirectional. When sounds hit the front of a bidirectional mic's diaphragm, it sends a positive voltage; the mic sends a negative voltage when sounds hit the back of the diaphragm. When using a figure-8 mic in an M-S setup, point the front of the mic to the left, so that one of the null spots is aimed forward, toward the sound source (see Fig. 1). This position allows the figure-8 mic to pick up ambient and reverberant information coming from the sides of the soundstage, although the sides will have opposing polarities. Note that the bidirectional mic is not a stereo mic: if you turn the middle mic off, you will not hear independent left and right channels from the side mic.
FIG. 2: Typically, directional and bidirectional mics are placed on top of each other for M-S recording.
Like the XY and Blumlein techniques, M-S is a coincident stereo technique. The two mics should be as close together as possible — they are usually placed one on top of the other — to avoid any phase-related coloration (see Fig. 2). In many coincident setups, much of the soundstage is picked up off-axis, where a directional mic may not sound its best, due to mic position. With an M-S setup, each mic is optimally positioned for its polar pattern and function. A stereo mic that gives you both a directional and figure-8 pattern, such as the AKG C 426, makes M-S miking easy. The Shure VP-88, a stereo mic that includes an internal M-S matrix, makes it even more convenient.
The signal from each microphone is then recorded to its own track. For you to hear a proper stereo image when listening to the recording, the tracks need to be matrixed and decoded. (I will explain how these processes work in a moment.) Otherwise, you will hear the direct image in one ear and the room sound in the other.
Do the Math
Although you have recorded only two channels of audio — the mid and the side — the next step is to split the mono side channel into two separate channels. This gives you three channels — the mid signal (center channel) with the two sides — that you can balance to re-create a stereo image. For M-S to work properly, however, all three channels must be present in some amount.
FIG. 3: M-S decoding requires a sum-and-difference matrix that adds one of the side signals to the mid signal to get the sum, and subtracts the other side signal from the mid signal to get the difference.
M-S decoding requires a sum-and-difference matrix (see Fig. 3), where you add one side signal to the mid signal to get the sum, and subtract the other side signal from the mid signal to get the difference. To do this, one of the side channels is shifted 180 degrees in polarity: when the polarity-shifted signal is added to the mid signal, you get the difference between the two signals. Using M for mid, S for side, and -S for the polarity-shifted side, the math looks like this:
M + S = left channel
M + (-S) = right channel
In other words, when two positive signals are added, you get a reinforcement of the signal — just like 1 + 1 = 2. When a positive signal is added to a negative signal, the signal is reduced — for example, 1 + -1 = 0.
In M-S decoding, the left bus of your stereo mixer sums the signal from the mid mic (panned center) and side mic (panned hard left), while the right mix bus sums the mid mic to the polarity-shifted mic (panned hard right). The concept can be further clarified, albeit in an overly simplified manner, by studying how the mics treat sound coming from three different directions.
FIGS. 4a–c: a. An instrument in the center of the stage makes a sound at 0º, which enters the cardioid on-axis and hits the null spot of the bidirectional. b. An instrument makes a sound at 45º to the left, which enters both mics. Because the positive lobe of the side mic is facing left, the side mic has a positive output, as does the cardioid. On the right side, the positive polarity from the side mic has been flipped to negative. When it combines with the mid mic, the signal is reduced. c. An instrument makes a sound at 45º to the right, which enters both mics. Because the negative lobe of the side mic is facing right, the side mic''s output has a negative polarity, while the cardioid mic is positive. In the side output with the reversed polarity, the negative signal is flipped and becomes positive. Then it is reinforced when it sums with the positive voltage from the mid mic. On the channel panned left, the voltage from the side mic remains negative. When it combines with the positive mid mic, the signal is reduced.
In Fig. 4a, an instrument in the center of the stage makes a sound at 0 degrees, which enters the cardioid mic on-axis but hits the null spot of the bidirectional mic. Because the cardioid is panned center, the signal is sent equally to the left and right mixer buses and speakers, giving you a centralized image.
In Fig. 4b, an instrument makes a sound at 45 degrees to the left. Because the front of the side mic is facing left, the mic's output has a positive polarity. The side mic's positive polarity adds with the positive polarity from the mid mic in the left mixer bus, resulting in imaging on the left side of the reproduced soundstage.
Over on the right side of the mix bus, the positive polarity from the side mic gets flipped to negative. When it is summed with the mid mic, the two signals with opposite polarity are combined, causing the sound level to be reduced.
In Fig. 4c, an instrument makes a sound at 45 degrees to the right, which also enters both mics. Because the negative lobe of the side mic is facing right, the side mic has an output with negative polarity. In the mixer, the negative signal is flipped in the channel that is panned to the right, making it positive. When summed in the mix bus with the positive polarity of the mid mic, it is reinforced and the sound is localized on the right. In the left channel, the negative polarity from the side mic is summed with the positive polarity of the mid mic, causing the sound level to be reduced.
Of course, in the real world the behavior of sound is a lot more complex when you consider frequency, room reflection, and phase. But these examples illustrate that, with the M-S technique, sounds are localized by differences in intensity at the microphones and by how the channels interact with each other electrically.
You can easily set up a matrix to decode M-S signals using a mixer. Some preamps, however, offer built-in M-S decoding, such as the M-Audio Octane and the Grace Design Lunatec V3. Standalone decoders are also available. Both types of devices usually accept mic- or line-level input from mid and side mics, matrix them internally, and give you a left/right stereo output. Typically, controls for mid and side levels are available on the box. A dedicated M-S decoder is especially handy in a remote- or a field-recording rig, because it allows you to monitor your signals without a mixer.
FIG. 5: An example DAW mixer set up to separate a finished left/right mix into M-S, adjust levels, and decode back to left/right. The Trim plug-ins are used for polarity reversal.
Although the signal path will vary depending on your equipment, the following is a generalized setup for recording and matrixing/decoding in M-S, using the mixer in Digidesign Pro Tools LE (see Fig. 5 and Web Clip 1). Begin by setting up the cardioid and bidirectional mics at 90 degrees to each other. Arrange your microphones so that the cardioid is pointing directly at the center of the sound source, and the front, or positive, lobe of the bidirectional mic is pointing to the left, as in Fig. 1.
Next, split the output of the bidirectional mic and assign each signal to its own mixer input. Ideally, that is done at line level, if you have outboard mic preamps, but it can be done at mic level as well. The split can be done in the analog world by making an XLR splitter cable, or in the digital world by assigning the same input to two different channels, either in a DAW or in a digital console.
In the mixer, pan the cardioid mic to the center and pan the two channels of your bidirectional mic in opposite directions — hard left and hard right. Invert the polarity of the bidirectional channel that is panned to the right. Many consoles and mic preamps have a button labeled Ø that does that. If not, you can easily make a short adapter cable that changes the polarity by reversing the hot (XLR pin 2) and cold (XLR pin 3) wires on the input end. (Be sure to clearly label the cable if you do this, so that you don't mistakenly use it later.)
In a DAW, you can use a plug-in to reverse the polarity of the signal, which I did in the Pro Tools mixer in this example using an RTAS plug-in. Now, assign all three channels to the mixer's main stereo bus, and your M-S decoder is complete.
Once you finish recording, take a moment to explore the stereo image. Bring up just the cardioid channel on the mixer. It should contain more direct sound than room reflections.
While dropping the level of the cardioid, bring up the two bidirectional channels together. As you do that, you'll notice that the stereo image gets dramatically wider as the center sound moves farther away. Eventually, when the mid fader is completely down and the side faders are up, you'll hear a wide, out-of-polarity signal, made up mostly of room reflections. To verify that everything is connected and working correctly, you can pan the two side channels to the center, and they should cancel each other out.
Neither of these extremes is an ideal stereo image, but they illustrate the different role that each mic plays in an M-S recording setup. By starting with the mid mic and mixing in the sides, you can find the right balance between the direct and reflected sound, as well as a nice stereo spread (see Web Clips 2 and 3).
As I noted earlier, one of the benefits of the M-S technique is that you can make these adjustments at any time after the recording is done, without adding extra processing, such as EQ, delay, or spatialization. Consequently, M-S recording is great for situations where fine-tuning the mic setup is not practical, such as when you're field recording. And because the decoding process can happen at any time, M-S is a popular choice when monitoring in less-than-ideal situations. Simply record the mid and side tracks to the two tracks of your stereo recorder, and balance them later using your matrix and decoder.
Back to Mono
As I mentioned earlier, an added benefit of M-S is perfect mono compatibility, because the stereo signal collapses to mono when the two bidirectional channels of opposite polarity are summed. In addition, the amount of reverb will also be decreased, which can add clarity when you are reducing a mix to mono. Even though other coincident techniques are theoretically mono-compatible, phase problems can still occur if the mics are not perfectly positioned. That is one reason why M-S has been used in broadcast recording for so many years.
An M-S setup also provides more options than other stereo techniques when it comes to microphone selection, because it does not require a matched pair. This gives you the freedom to choose different mics for the mid and side without negatively affecting the stereo image.
For example, in a particular situation you may prefer the clarity of a condenser for the mid mic, but the body and warmth of a ribbon for the side reflections. That kind of flexibility is great for solo instruments large and small (such as acoustic guitar and drum overheads). With large ensembles, however, which have a wider stereo image, the side mic may capture a significant amount of direct sound from the instruments. In that case, matched mid and side microphones are a good choice, in order to keep the stereo image even and stable.
Classical and jazz engineer Steve Bellamy, who is based at the Banff Centre in Canada, says he prefers M-S “on acoustic guitar and on piano recordings when I want to be close. When the source is large, like a piano, the M and S mics should be matched because they are sharing panning information. For smaller sources, like guitar, you can color the wider, spacious sounds differently than the center image simply by choosing a brighter side mic.”
Everyone's a Critic
Of course, every recording technique has its trade-offs, and M-S is no exception. It is important to listen to how your recordings sound when using the mid-side setup technique, to make sure it is the right choice for the situation. If the microphones are not positioned well, no amount of tweaking will make them sound better. The best idea is to fix any problems before you record.
Begin by listening for sonic artifacts such as an imbalance between the mid and sides. If there is too much mid, the image will collapse to near-mono; if there is too much side, the image will sound unnaturally wide, unfocused, and phasey. Having to fight with the mix to get a good tone, balance, and imaging may mean that M-S is not the right choice for that recording situation.
Like other coincident techniques, M-S has been criticized for lacking the spaciousness that a spaced pair of omnidirectional mics offers. Although there are trade-offs with spaciousness (primarily in precise imaging), there are techniques that pros use to make the imaging of M-S more dramatic. For example, you can use an omni as the mid mic instead of a cardioid. That allows you to have the control that M-S offers, with the spaciousness of an omnidirectional mic. Using an omni in the center also gives you the benefit of an extended low-frequency response, which cardioids don't typically offer.
You can use any polar pattern you want for the mid mic — hypercardioid, omnidirectional, or bidirectional. Many engineers supplement a center M-S setup with an additional pair of omni mics further out in the room. That allows them to use the M-S setup for focus and the omnis for added spaciousness and width.
As part of his research in developing the Lexicon reverbs, David Griesinger suggests that the spaciousness of M-S can be increased with what he calls a “spatial EQ.” Because low frequencies have been found to increase the level of the listener's envelopment in the sound (particularly when they are different in each channel), he recommends boosting signals below 400 Hz by 2 to 4 dB in each of the side channels, often with a corresponding cut in the mid channel.
Dial M-S for Mastering
M-S is not only a recording technique, but also a way to process stereo audio. As a result, it's one of the mastering engineer's secret weapons.
M-S lets you separate the sounds in the center, which are common to the left and right channels, from the sounds on the sides, which do not share information with each other. For example, that allows you to change the balance of the final mix by encoding a pair of left/right stereo tracks to M-S tracks, processing or adjusting them independently, and decoding the altered M-S tracks back to left/right stereo.
How the M-S processing is used varies with the material at hand. The most common application is to change the width of the stereo image, either collapsing it for mixes that are too wide by increasing the volume level of the mid, or widening narrow mixes by increasing the level of the sides. That technique works best in small doses, because a little can go a long way.
Once the tracks are isolated as mid and side components, they will each contain different instruments, which can be used to your advantage. For example, you can brighten up a lead vocal in the center without affecting the stereo percussion on the sides (see Web Clip 4); move a solo piano a little closer in the mix by minimizing the reverb from the hall; or add a small amount of compression to tighten up the rhythm-section elements in the middle, while not adding audible artifacts to the cymbals at either side.
Some mastering processors have built-in M-S processing capabilities. For example, the TC Electronic System 6000's BackDrop broadband noise reduction allows you to take a noise print and work on it in either left/right or M-S, automatically doing the encoding and decoding for you. With processes as program-dependent and touchy as noise reduction, the M-S option can make all the difference.
“M-S processing can help you control aspects of the mix that are hard to reach with standard left/right processing,” says Greg Reierson, principal of the Minneapolis-based Rare Form Mastering. “The most obvious is widening or narrowing a mix, but other image-specific issues arise. I sometimes find I want to adjust frequency-specific content of the side channel without affecting the center, or vice versa. If done carefully, M-S allows more specific control of one channel without adversely impacting the other.”
Several DAW plug-ins are available that convert M-S sources to left/right stereo and vice versa. That can save channels and mitigate error in DAW routing. A plug-in like the Waves S1 offers M-S processing — with and without spatial EQ — as part of its spatialization algorithms.
Be sure to listen carefully when making these kinds of mastering changes. There can be a tendency to overdo the side levels in hopes of creating an impressively wide image. Not only does this change the musical balance, but it could also leave you with phasey-sounding sides and a weak center.
Getting There and Back Again
The process for encoding left/right to M-S and decoding it back again requires a few more channels than M-S recording, but the tools and concepts are the same (see Web Clip 5). The goal is to separate everything that the left and right channels have in common from what is unique to one channel. That means reversing the sum-and-difference matrix used for recording.
Begin by making two copies of the original left/right mix. Duplicate them onto four mono tracks on a DAW or split them to four channels on the console.
Pan the left/right tracks from one copy of the mix to the center. If you like, send them to a mono subgroup in order to control them with one fader. That will become your mid, or sum, channel.
Next, pan the tracks from the other copy of the mix to the center, and reverse the polarity of the right channel. Send them to a mono subgroup, too, if you like. That will become your side, or difference, channel.
Now you have independent mid and side channels from a left/right mix, which can be balanced and processed independently. Because you'll want to listen in left/right stereo, take the mid and side channels and decode them as if they were M-S mics.
What you've done is add the two left/right channels together to make a mono signal that creates the mid channel. The side channel was also created by adding the left/right channels together. When the polarity of one of the channels is reversed, however, any sound that was common to both gets canceled.
You can prove the math by taking a left/right pair, encoding them as M-S, and then decoding back to left/right without making any changes. Assuming your gear is operating as it should, the original and post — M-S left/right tracks will be exactly the same. (To be sure, try flipping one pair out of polarity against the other.)
These are just the basics of the mid-side technique. If you want to get deeper into the technique, read The New Stereo Soundbook, by Ron Streicher and Alton Everest (www.stereosoundbook.com). It goes further into the details and is an excellent resource on microphones and stereo sound in general.
The best way to become adept at the mid-side technique is to experiment with it for recording and mastering. Armed with a bit of confidence in what M-S can do, you will have another powerful tool in your arsenal, and a new way to think about stereo.
Brian Heller (www.whattheheller.com) is a composer, engineer, and tech based in Minneapolis. He teaches in the Sound Arts program at Minneapolis College but is currently staff engineer at the Banff Centre, in the Canadian Rockies. He would like to thank Greg Reierson of Rare Form Mastering and Steve Bellamy of the Banff Centre for their feedback.