Part II

Recording
Chapter 3

Basic two-microphone stereo techniques

The focus in this chapter will be on the two-microphone stereo techniques

that are frequently referred to throughout the book for non-orchestral
recordings such as piano, chamber music, choirs, and organs. These
fundamentally useful techniques are co-incident directional microphones,
spaced omnis, and spaced and angled cardioids (of which the ORTF pair is
the most well-known example.)

3.1 Co-incident microphone techniques

When two microphones are mounted as close together as possible (or when
a dedicated stereo microphone with two capsules is used), the microphones
are said to be co-incident, meaning that they are located at the same point
in space. In theory, there are no timing differences between the two signals
generated by such a pair, as sound waves from any source will arrive at
both microphones at the same time. The resulting stereo image is produced
by level differences between the two signals; these level differences are
created naturally by the use of directional microphones pointing in different
directions. Given that it is not really possible to mount two microphones in
exactly the same place, there will be some very small timing (and therefore
phase) differences, but these can be disregarded in practice.

To make the terminology clear, Figure 3.1 shows a general co-incident pair
of unspecified directivity pattern (the patterns most used will be fig of 8,
cardioid, and hyper-cardioid). The microphones are mounted with an angle
between their front axes, and depending on this mounting angle and the
directivity pattern, the pair will have a characteristic L-R ‘stereophonic
recording angle’. This is the angle between the positions of sources that
will appear fully left and right in the stereo image produced by the pair, and
it is not the same as the mounting angle. (It is assumed that the pair of
microphones are fully panned.)
The L-R stereo recording angle is dictated by the level differences between
the microphone signals that arise due to the directivity of the pair, and an
individual instrument will only be perceived to be fully left or right when
the level difference between its signal on each microphone reaches about
18 dB.1 The width of the final image of the source will depend on how
much of the pair’s stereo recording angle it occupies. If the pair being used
has a very wide stereo recording angle, and the source is a piano that only
occupies a small segment of it, the recorded piano image will be narrow
even with the pair fully panned. To make the piano wider without moving
closer, the L-R stereo recording angle needs to be reduced so that the piano
occupies more of it. In general, an increase in the mounting angle results in
a decrease of the stereo recording angle of the pair. This means that
increasing the mounting angle will also increase the image width of the
source.

Figure 3.1

Figure 3.1 Generic co-incident pair

The most well-documented co-incident microphone technique is the

Blumlein pair, which consists of a pair of fig of 8 microphones mounted
with their front axes 90° apart. This pair has a stereo recording angle that is
also 90°, but care must be taken not to place any sources outside this angle
to the sides of the pair, as the source will be picked up on the front lobe of
one microphone and the rear lobe of the other, meaning that one signal will
be phase inverted. When reproduced over left and right speakers or
headphones, this phase inversion is a very unpleasant listening experience.
Co-incident fig of 8 microphones are almost invariably used mounted at
90°.

When cardioids are used as a co-incident pair, the problem of a phase-

inverted region does not arise, but some mounting angles will produce
significant angular and level distortion in the stereo image. Angular
distortion means that some parts of the image are compressed into a narrow
arc, and other parts are stretched out to occupy a wide section of the image.
This occurs because of the way that the level difference between the
microphones changes with the angle of incidence. Level distortion means
that some parts of the stereo image are louder than others (the areas of the
image that are compressed into a narrow range are also louder.) The
practical upshot is that cardioids mounted at 90° will produce an image
which is centre heavy; most of the image is squashed into a narrow central
range, with the rest stretched out at the sides. Conversely, cardioids
mounted at 180° will produce an image that is stretched out and quieter
across the centre and squashed up at the sides. If co-incident cardioids are
to be used as a main pickup, they should be mounted at an angle of about
135° as this produces a useful L-R stereo recording angle and the least
angular and level distortion. However, cardioids mounted at 90° can be
useful for placing on a single, smaller instrument for use as an ancillary
pickup. They will produce a very phase-coherent, centrally focussed image.
Figure 3.2 shows the mounting and approximate stereo recording angles of
cardioids at 90° and 135°.

Figure 3.2a-b

Figure 3.2a–b Cardioids mounted at 90° and 135°, respectively

Co-incident pairs create an image that is stable and coherent but lacks a
feeling of spaciousness, and because directional microphones are used, the
fullest LF range is not captured.

3.2 Spaced omnis

Using spaced omnis as an overall pair has both positive and negative
characteristics. In its favour, the tonality will be excellent, with full LF
range being included, and there will be a spacious feeling to the recording
partly due to collection of more room sound, and partly because of the
phase differences between the left and right channels. The downside is that
the image will not be sharply focussed or precisely located, and if the
microphones are too far apart, the centre of the recorded image will be low
in level and the outer edges will dominate (the ‘hole in the middle’ effect).
The lack of image precision can work to the engineer’s favour if spot
microphones are going to be extensively used; it means that they can be
panned in a wider range of positions without causing conflict with the
spaced omni pair. (See Chapter 13 for recording the classical brass
ensemble as an example.)

The microphones will need to be mounted on a wide stereo bar (see

Chapter 2), and it is common practice to angle them outwards at about 30°
to 45° to take advantage of any HF directionality in the microphones. The
stereo image when using spaced microphones as an overall pickup is
predominantly created by timing differences between the two signals,
although there will be some level differences at HF due to the HF
directionality of omni microphones if they are angled outwards. The
resultant source image width is therefore dependent on the spacing between
the microphones, and this must be adjusted to take into account the size of
the instrument or ensemble, its distance away, and the desired recorded
image width. (See also section 5.5.2.) Spacing between fully panned omnis
reaches a natural limit at about 1 m (3′4″), and beyond this, the timing
differences start to become too large and there is not enough correlation
between the microphone signals to create a stereo image. Reproduction of a
large orchestra using spaced omnis is best addressed by use of the three-,
four-, or five-microphone Decca Tree (see Chapter 8).

A small table of typical spacings used in practice to produce a conventional

image width is included in Figure 3.3. These should be taken as a good
starting point from which to adjust position and spacing for image width
and amount of direct sound versus reverb. To make the sound less
reverberant, the microphones should be moved closer to the source, bearing
in mind that this might also increase the image width by virtue of being
closer to the players, so some microphone spacing adjustment might also be
needed.

Figure 3.3

Figure 3.3 Table of typical omnidirectional microphone spacings

As noted in Chapter 5, spacing greater than about 33 cm (13″) on piano

starts to produce an over-wide image that has poor localisation. Conversely,
if a 30 cm (12″) spaced pair is used on an orchestra and fully panned, the
image will still be too narrow to fill the space between the loudspeakers.
The microphones should be panned fully left and right (or very close to it).
Panning a main pickup inwards to reduce the image width of the source has
the effect of narrowing the sense of space and reverb around the
instruments as well, and this produces a rather mono-sounding recording.
Additionally, because this technique involves the creation of timing
differences (and therefore phase differences) between the two microphones,
panning them inwards to any significant degree (less than about 70% left
and right) will start to produce some colouration due to partial summation
of signals of different phase. When using spaced omnis as a main pickup
(rather than for a ‘stereo spot’ as discussed in Chapters 6 and 7), it is best to
space the microphones a little less if the image width needs to be reduced
and to pan them fully.

3.3 Spaced and angled cardioids

A very useful technique that combines some aspects of co-incident pairs
and spaced omnis is the use of cardioids that are spaced a small distance
apart and angled away from each other. This general technique produces a
lot of possible combinations of angling and spacing that might prove useful
(in that they produce a useful L-R stereo recording angle and no significant
angular distortion); these were quantified by Michael Williams2 in 1984.
The stereo image is created by a combination of level differences (due to
the directionality of the microphones) and timing differences (due to the
spacing between the microphones). The resultant image is more focussed
than the spaced omnis, but with a more spacious feel than a co-incident
pair. Where the extended LF range of omnis is not required (such as for a
choir), they make a very useful overall pickup. As with the spaced omnis,
the pair should be panned fully left and right as the default, and altering the
angle or spacing would be preferable to panning them inwards beyond
about 75% left and right.

Of all the possible spaced and angled pairs, two have passed into common
use: the ORTF (Office de Radiodiffusion Télévision Française) and NOS
(Nederlandsche Omroep Stichting) pairs, as shown in Figure 3.4.

It can be seen that the NOS pair has wider spacing and narrower angling
than the ORTF pair, indicative of the trade-off between the two aspects of
this sort of pair. As we saw in section 3.1 (co-incident pairs), a wider
mounting angle means a wider image, and from section 3.2 (spaced omnis),
wider spacing produces a wider image. The ORTF-type technique can be
used flexibly in practice, and the angling or spacing can be altered a little to
change the image width if changing the distance from the source is not an
option. This might arise because the balance between direct and reverberant
sound works well at a particular distance, or because the rig cannot be
moved elsewhere because of space restrictions.

Figure 3.4a-b

Figure 3.4a–b ORTF and NOS pairs

Elsewhere in this book, parallel-mounted pairs of spaced cardioids have

been used extensively as spot microphones on soloists. This technique
could be seen as part of the family of spaced and angled pairs as they have
a mounting angle of 0° and spacing in the 20–30 cm (8″ to 12″) range.
These ‘stereo spot microphones’ produce an image of the soloist that is
wider than a single mono spot but is still appropriately restricted in width
when fully panned.

Notes
1See Rumsey & McCormick: Sound And Recording
(Focal Press ISBN-13: 978–0240521633)
7th Edition p498 for a useful summary of the research that underpins this
finding.
2The Stereophonic Zoom: A Practical Approach to Determining the
Characteristics of a Spaced Pair of Directional Microphones
Author: Williams, Michael
AES Convention: 75 (March 1984) Paper Number: 2072
Publication Date: March 1, 1984
Subject: Studio Technology
Permalink: www.aes.org/e-lib/browse.cfm?elib=11692
There are interactive displays of stereo recording angles for various
stereo techniques at www.sengpielaudio.com/Fragen08.htm
Chapter 4

Solo instruments

This chapter aims to cover those instruments for which there is a large body
of unaccompanied classical repertoire. This includes both polyphonic
instruments that provide their own harmonies (such as the guitar, harp,
harpsichord, and other early keyboard instruments) and instruments that
primarily produce a single line (such as violin, cello, and woodwinds). For
instruments that are almost always employed as part of an ensemble, it is
more practical and useful to consider how to record the whole ensemble.
Therefore, discussion of brass instruments has been left to Chapters 7 and
13.

Throughout this chapter, the discussion is concerned with placement of spot

microphones on a single player. It is assumed that in addition to the spot
microphones, there will be some sort of overall room pickup to create a
good backdrop of stereo, reverberant sound into which the spot
microphones can be blended. Unless you are lucky enough to be working in
an excellent acoustic environment, using two pairs will be needed to
produce a sound that works well in all aspects. The overall room pair could
be an ORTF-type, or omnis spaced at around 60 cm (2′), at about 2.5–3 m
(8′ to 10′) high and set back by 2.5–4 m (8′ to 13′), depending on the
acoustic.

When recording a solo instrument, the width of the recorded image of the
instrument is a primary consideration. The sense of space and reverb
around the instrument should fill the image width between the loudspeakers
completely, but the width of the instrument within this should not fill the
stereo picture if it is to give a realistic illusion of a live performance. A
single microphone placed appropriately will be able to capture a good tonal
overview of an instrument as long as it is not too close, but the result can be
very one-dimensional unless there is also some sort of stereo pickup
contributing significantly to the instrument’s sound. This aspect of imaging
as it relates to a solo singer is discussed in Chapter 6, and the same
principles can be applied to solo instruments. Even with additional stereo
artificial reverb, or stereo ambient pickup microphones, a single spot
microphone can dominate the mix, and the instruments’ image may become
too narrowly localised, as if listening ‘down a tunnel’. Therefore, the use of
two spot microphones for solo instruments is something to be seriously
considered, to create a sense of width and ‘bloom’ around the image. The
desired lateral image width of a solo harpsichord will be a greater than that
of a classical guitar, which will be a little greater than that of a solo violin
or oboe, but none of them should be expanded to fill the whole width of the
stereo field or collapsed to a mono point. There is, of course, a range of
image widths that will produce a great-sounding illusion of a solo
instrument playing in a real space, and the final judgement is for the
engineer to make. A narrow image can make the listener feel further away,
even if the microphones give a sense of being closer. For the sense of
distance to work, there should be no conflict between the width of the
image, the close detail of the sound, and the amount of reverb.

4.1 Classical guitar and flamenco guitar

These instruments are very similar to one another, both being nylon-strung
(unlike the steel-strung acoustic guitar), and the recording approach can be
the same.

Guitars, lutes, and theorbos are all quiet instruments with a restricted
dynamic range, so the recording venue should have a very low level of
background acoustic noise, and the microphones should have low levels of
self-noise. Occasionally, classical guitarists will use a small amplifier with
a freestanding microphone if performing solo or a concerto in a very large
venue that is better suited to orchestral music. This amplifier is not an
integral part of the sound; it is there to make it louder as transparently as
possible.

The sustained sound from the guitar comes primarily from the resonating
front body; the string vibrations are transmitted to the body which acts as
an acoustic amplifier. The sound hole is not the main source of the sound,
although it is a tempting target at which to aim microphones. From the
point of view of the recording engineer, the next most prominent features of
guitar playing are plucking sounds from the right hand, and noises and
squeaks from the fingerboard. Players always try to keep these noises from
left-hand position changes to a minimum, but will vary in their ability to do
so. They are usually very concerned with noises not being prominent in the
recording, and you are likely to have to spend some time trying to minimise
them at the time of recording (the best option) or painstakingly removing
them during post production (soul-destroying, time-consuming, and
therefore expensive). In flamenco playing, striking the front of the guitar
(golpe) is part of the technique, and this can produce a large LF resonance
from the whole body of the instrument as well as a sharp transient at the
point of impact. Use of this important percussive technique will increase
the overall acoustic level when compared with a purely classical guitar, and
might also be used in more contemporary classical repertoire. Other player
noises such as stomach rumbles can become obtrusive when amplified
alongside a low level guitar signal.

4.1.1 Microphone placement and image

There is a lot of material written about pop microphone placement for

acoustic guitar, but they are generally placed too close to the instrument for
a classical-style solo recording. The emphasis is often on two separate spot
microphones placed around 20–30 cm (8″ to 12″) away from the
instrument, each capturing a certain aspect of the sound such as the main
body and the fretboard. These microphones are usually too far apart to form
a coherent enough stereo pair, and so they form two pools of mono signal
that are combined to create an overall tonal representation of the
instrument, if not a good spatial impression. They may be panned almost to
the same place, but the aim in a pop song is not to create the sense of a
guitar playing in a real space but to capture the closer sound of the
instrument that can be placed into a mix.

For classical guitar, a simple narrowly spaced stereo pair can be placed at
about the height of the centre of the guitar, perpendicular to its front
surface. If the guitar is angled upwards slightly when played, move the
microphones higher to look down a little. Fine tuning the placement of the
microphones can be helped by sitting on the floor with your head at
microphone height, and moving around the instrument to see how the
sound changes. As a starting position, aim the left microphone at the
bottom of the body and the right microphone at the bottom of the fretboard.
This will place the fret noise a little to the right, but this seems to work
acceptably when listening to a guitar recording. Lateral movement around
an arc will help to reduce or increase the noise from the fretboard if it is too
intrusive. (See Figure 4.1.)

The pair should be spaced around 25 cm (10″), pointing forwards at a

distance of 70–100 cm (2′4″ to 3′4″) from the guitar (if the room is a nice
sounding one, a little closer if not) with a suitable amount of natural,
smooth reverb. The image width that works best for classical guitar is to fill
around 30% of the total stereo field. If it were too wide, it would give the
listener an unreal sense of the instrument’s size, and the creation of a real
sense of perspective and space would not be successful. A guitar would
only fill the listener’s field of view if they were seated about 30 cm (1′)
away, but the microphone placement and amount of reverb used give cues
to the listener that place the player further back than this.

Finding an appropriate balance between the reverberant and direct sound by

altering the distance at which the microphones are placed can be difficult to
get right with classical guitar. There is a conflict between trying to pick up
this quiet instrument in a reverberant environment by moving in closer, and
avoiding the close microphone placement that produces intrusive fret noise
or a very localised tone quality. If you have problematic fret noise, move
the microphones to at least 1 m (3′4″) away, and adjust their position
around the arc by ear (see Figure 4.1). If you find that omnidirectional
microphones at this distance are now picking up too much reverb, switch to
directional microphones. This will cut out some of the reverb and give you
a dryer but not too close sound to work with. Ribbon microphones (Royer
R-121 or Coles 4038) in particular can be useful because their HF roll-off
will soften fret noise, and an additional, more ambient pair of omnis can be
used as a room pair to collect some reverb and warmth of tone.
 Figure 4.1a Classical guitar microphones – plan view showing
line of lateral adjustment

Figure 4.1b

Figure 4.1b Classical guitar microphones – side view

A room with a high level of background noise will present similar, related
difficulties. Microphones need to be close to record sufficient level from
the instrument over the background noise, but this means that you might
struggle with prominent finger squeaks and lack of reverb. Additional room
microphones will be needed for the room reverb, and using ribbon
microphones might help with the squeaks.

If you have to record a guitar, lute or theorbo in an orchestral or large

ensemble context, you will need to be closer to pick up sufficient level and
detail. This will of course be in addition to an overall orchestral pickup, so
will be primarily used to make sure the transient details are not lost, and
some contact with the finger noise is actually desirable to help the
instrument be heard. A distance in the order of 15–25 cm (6″ to 10″) would
be recommended, but you will need to beware of the possibility of
proximity effect (an increase in the low frequency components) on any
directional microphones at this distance, and correct for any excessive LF
colouration if it is audible once mixed with the orchestra. If you have to use
a single microphone in this context, start with it about halfway along the
total length of the guitar (neck and body combined) and move it laterally
until the balance between fretboard and main body of the sound is
satisfactory. (See also Chapter 11 for concerto recording.)

4.1.2 Using artificial reverb on guitar

If the space in which you are recording is a dead room, or has poor
sounding reverb, then artificial reverb can be added afterwards instead.
However, artificial reverb can inadvertently emphasise finger noise that
was relatively unobtrusive when it was recorded dry. A squeak is much
more troublesome to the listener if it has 2–3 seconds of bright sounding
reverb tail attached. In a real space, the HF content of the squeaks does not
travel too far into the room because high frequencies are partly absorbed by
the air. Hence, the ratio of ‘squeak to music’ picked up by the microphones
gets higher the closer to the instrument the microphones are. The squeaks
are not as present in the real room reverb as the other parts of the sound
because they are at a reduced level by the time they reach a room boundary
for reflection. If you are adding artificial reverb, this is something that you
can try and control as part of setting the parameters. If you are using a
programme that does not enable you to control the HF behaviour of the
reverb algorithm in a sophisticated way, you can use a gentle HF roll-off on
the auxiliary send that you are using to drive the reverb, and a similar roll-
off on the reverb returns. Adding a small amount of EQ to both send and
return will often give better results than EQ’ing only one of them more
drastically.

4.1.3 Microphone choice

The lowest note on the classical guitar is E2 at approximately 82 Hz, and so
some good cardioids (e.g. Neumann KM84) will adequately cover the
frequency range without the loss of any fundamental frequencies. Omnis
(such as the KM83) might be preferred for the main guitar microphones if
the location is good, and a natural balance between the room and the
instrument can be found. Because they have a more extended bass end, you
are likely to find that they will pick up some lower resonances from the
body, but the preference for the sound is a personal one; the use of
directional microphones should not affect important frequency content
unless they have a roll-off that affects the 50–100 Hz range. As noted
earlier, ribbon microphones will work very well, and they have the
advantage that the HF roll-off will reduce the prominence of any fret noises
and squeaks which can be particularly useful if you have to get in a bit
closer. With fig of 8 ribbon microphones in particular, you should be aware
of the possibility of the proximity effect starting to cause some artificial
bass colouration if you are closer than 60 cm (2′) or so. This can be
attractive in small amounts, but LF is seductive; you should be careful of
being seduced by a sound that is large, warm and boomy but not actually
like a classical guitar. As discussed earlier, using omnidirectional
microphones for an ambient room pair, and ribbon microphones or
cardioids for a more detailed pickup of the guitar can be a very successful
combination. Avoid microphones that have any sort of HF lift above about
7 kHz that can make the instrument sound thin. If electrical noise (hiss) is
proving a problem and you have optimised the rest of your signal path and
gain structures, it is worth remembering that large diaphragm condenser
microphones have a lower self-noise and higher sensitivity than those with
small diaphragms, so require less mic gain.

4.1.4 Floor and surfaces

Floor reflections from a wooden floor can really enhance the sound of a
classical guitar. Stone floors are too bright and immediate, as are some
types of modern varnish on a wooden floor. If you have to record in a
carpeted space, try placing some sheets of wood, plywood or medium-
density fibreboard (MDF) under the player’s chair to improve the sound of
the instrument.
4.2 Harp
The orchestral concert harp has a wide dynamic range, and it is a large and
powerful instrument when compared with its folk music cousins. Its sound
consists of a distinctive plucked transient followed by a sustained
resonating tone that comes both from the strings and the soundboard
(although this is more of a resonator like a guitar body than a soundboard
like that of a piano.) The range of its strings’ fundamental frequencies
extends from C1 at approximately 32 Hz to G7 at 3.1 kHz, so it approaches
the piano in terms of its range once overtones are included. Therefore, in
order to pick up the full effect of the lower frequencies in a solo recording,
using omnidirectional microphones to form at least part of the sound is to
be recommended.

The concert harp also has pedals which are manipulated to raise and lower
the pitch of a given set of strings, creating the ability to play sharps and
flats as required, and making the instrument chromatic in nature. These
pedals will make some noise when they are being engaged or disengaged,
but as long as they are not too obtrusive in the recording, they can be
considered as part of the instrument’s natural sound in the same way as
woodwind key noise and guitar fretboard sounds. The opinions of
engineers, producers, and performers concerning how much instrumental
noise is tolerable can vary widely; it is unrealistic to try to eliminate these
sources of noise completely, and accepting them as part of a real human
performance on a real instrument will avoid the recording being processed
into a state of unnatural and clinical quietness.

4.2.1 Solo harp image

The temptation when recording a solo harp might be to create an over-wide,

rather ‘technicolour’ harp so that a full glissando travels from one
loudspeaker and into the other. While this might have been fun in the early
days of stereo and film scoring, or if used as a magical sound effect, we are
trying to create the illusion of the sound of a real instrument in a real space,
and therefore we need to constrain the image to something more central. In
accordance with the techniques outlined for soloists so far, a forward facing
pair spaced at 25–30 cm (10″ to 12″) apart can be used to create an image
of the harp that is not overwide even when fully panned, although angling
the microphones apart a little will increase the image width if this is
desired. Spacing the microphone more widely will result in two less
correlated sources, one picking up lower strings, and one higher. This will
create some dramatically wide glissandi when panned left and right, but
little sense of a real instrument or a real space. (See Chapter 9 for dealing
with the harp in an orchestral context.)

4.2.2 Microphone position and choice

The player sits with the harp resting on their right shoulder with the
soundboard close to their body. The player can then look along the rows of
vertical strings, with the highest notes nearest to their right ear and right
hand. This allows the player to reach forward with the left hand to the
lower strings, and to have a clear view of the music stand which is placed
on the left-hand side of the instrument (from the player’s perspective).

Figure 4.2a and 4.2b show the harpist and some microphones positions to
be discussed below. Positions A and B are useful; position C is to be
avoided.

Figure 4.2a

Figure 4.2a Harp – plan view showing alternative microphone

positions A and B. Position C is not recommended

Figure 4.2b

Figure 4.2b Harp – end view showing height for positions A and
B

The microphones for positions A and B are placed sideways on to the

instrument, aimed at a position midway down the strings from a height of
around 150–180 cm (5′ to 6′) and about 90–120 cm (3′ to 4′) from the harp.
In position A, the microphones are on the player’s right-hand side, and this
can be preferable to position B, as the finger noise is a little reduced from
this side. In all other respects, position B will also work well, as long as the
microphones can be placed to look above the music stand. Omnidirectional
microphones would be the preferred choice if the room is sufficiently nice
sounding, because of the extended frequency range of the instrument. The
distance of 90–120 cm (3′ to 4′) from the harp will be close enough to get
some good detail and pick up the quietest passages from the top strings, but
not so close that the sound becomes localised to one part of the instrument.

If the engineer concentrates on the soundboard as the source of all the

sound and moves the microphones around in a 90° arc to look straight at
the soundboard from the front end of the harp (position C), odd phase
relationships and cancellations can result. A more extreme version of this
approach is to put one microphone on either side of the harp; this will result
in two rather uncorrelated microphone signals which will not create a
convincingly located instrument in the recording image. As noted at the
start of the chapter, an overall room pair should be added to complete the
sound.

4.2.3 Floor and surfaces and noises

As with the guitar, a nice wooden floor will enhance the resonance and
sound of the harp. However, the pedals will create some noise acoustically
and also have the potential for transmitting noise mechanically through a
sprung floor. The best way to ameliorate this is to use some good isolation
mounts for the microphones to avoid pedal noise transmitting itself in this
way. If you are still picking up too much pedal noise, have a listen in the
room with the player and see if this is how the instrument sounds anyway.
If it is noisy in real life, talk to the player about it to see if anything can be
done. If it cannot be improved by the player adjusting the instrument, it can
be worth adding a small piece of carpet under the pedals as you might for
piano, where the back of the player’s heel is on the floor. Avoid using a
large rug as it will be better for the overall sound for most of the interaction
with a wooden floor to be included. Remember that some pedal noise is
fine as it is part of the instrument; along with finger noise, it is only a
problem if it starts to sound very close and out of context with the rest of
the sound. This arises because we usually place microphones closer than
we would sit to a performer in a concert, and if misjudged, this can lead to
instrumental noises becoming too obtrusive.
4.3 Violin
The violin (along with the guitar and orchestral stringed instruments)
consists of resonating strings that are coupled to a body that acts as an
acoustic amplifier. The body produces the majority of the sound from its
own vibrations and that of the air inside it. It has a very extended frequency
spectrum containing high levels of HF overtones that give the instrument
its character. Both the higher frequencies (for the sense of contact and
connection with the instrument) and the lower frequencies (for a feeling of
warmth and resonance) are needed for a beautiful and well-balanced solo
violin sound.

The radiation pattern is complex, but taken as a simplified overview, the

lowest frequencies are radiated more or less omnidirectionally, and the
higher frequencies are beamed upwards perpendicularly to the front face of
the body.1 The higher the frequency, the narrower the beam of radiation.
Consequently, placing microphones above the player will capture all the
frequencies produced but may not be very flattering if the player is less
skilful or harsh in tone. In a concert situation in an appropriate
environment, the room will integrate the whole sound before it reaches the
listener, in a similar way as happens with a classical voice. When recording
a live concert, the choice of where to put microphones is restricted because
of the resulting visual intrusion; thus microphones are often placed much
lower down, and hence they do not capture the upper overtones. The
engineer is then reliant on capturing some of this HF on other microphones
within the room, usually on some sort of overall pickup placed higher and
further back.

4.3.1 Microphone placement

Violinists are like singers in that they will move around quite a significant
amount during a performance, and any microphone technique needs to be
able to manage this situation adequately, without the image moving in and
out of focus or from side to side. The majority of soloists will perform
standing up, so all distances given in this section are in relation to a
standing player of middling adult height, perhaps 1.7 m (5′7″). If the player
is seated, you will need to lower the microphones accordingly.

With a professional player with a good tone, a reasonable starting point

would be to place a pair of microphones about 90–100 cm (3′ to 3′4″) back
from the player, looking down at the instrument from a height of 2.4–2.85
m (8′ to 9′6″) (see Figure 4.3a and 4.3b). This should produce a well-
balanced overview of the whole sound, the aim being to achieve a good
balance between a sense of space around the instrument and giving the
listener a feeling of immediate emotional contact with the player. The
balance between contact and spaciousness can be altered with the height of
the pair, and it must be judged in a given context; the height that works will
depend on how reverberant the room is and the directivity pattern of the
microphones. The distance between the microphones and the violin also
serves other purposes; it will reduce any uncomfortable closeness and
intrusive bowing noise by means of the loss of HF with distance, and it will
reduce the impact of player’s movements on both the tonality of the sound
and the stability of the image. Microphones placed lower down in front of
the player at around the instrument’s height of perhaps 1.5 m (4′11″) will
lose the ‘presence’ of the bowing and HF components, and will need to be
further back than 1 m (3′4″) to recover an appropriate balance between
direct and reverberant sound within the room. Overall, it is harder to
achieve a really pleasing balance between spaciousness and detail with the
microphones placed lower down.

Figure 4.3a

Figure 4.3a Violin spot pair – plan view

 Figure 4.3b Violin spot pair – side view

Player movement is something that can cause problems, particularly when

using a pair of spot microphones. Initially, the microphone pair should be
25–30 cm (10–12″) apart, parallel to one another, and panned fully left and
right to produce a central image with some ‘bloom’ around it. If the image
swings around a lot when the player moves, these can be panned in up to
around 75% (following the same principles outlined in Chapter 6) to reduce
image instability. If this isn’t sufficient to tame the movement, you can
reduce the spacing of the pair to the lowest end of the suggested range, and
physically angle them inwards slightly. Another technique that works
extremely well is to add a central ribbon microphone (Royer R-121 or
Coles 4038) to the violin pair and use this at a lower level in the mix to
stabilise the image, as discussed in section 4.3.2.

Figure 4.3 shows a good starting position for placing the close pair on a
violin.

The most common problem will be too bright a tone, or too harsh a tone
with a less good player; this can be reduced by changing the microphones’
horizontal angle and pointing them a little over and beyond the instrument
(Figure 4.4a). This will move the instrument away from the front axes of
the microphones and thus reduce the HF content. If this is not sufficient,
you could reduce the height of the microphones in order to move them out
of the way of the highest frequency projection that is perpendicular to the
instrument’s bridge and front face (Figure 4.4b). However, the microphone
height does give a good feeling of space, and it can help avoid floor
reflections, so rather than sacrificing height, you can also try moving the
microphones around in an arc as shown in Figure 4.4c. The further around
towards the left (on the diagram) that you go, the more the HF will be
reduced as it will be effectively blocked by the player’s head. Microphones
behind the player might be useful in a concert scenario; however, they will
also result in a loss of HF as the player holds the instrument to project
forwards and upwards – not behind him or her. The best choice of
technique for altering the HF content might depend on how the player
characteristically moves; spend some time watching them play and observe
where they tend to move and how much they move. This should enable you
to get a good microphone position according to their ‘average’ playing
position. If movement is extreme, microphone placement off-axis from the
instrument’s bridge will reduce the degree of HF change that occurs as the
instrument swings around. (Figure 4.4a through 4.4c show different ways to
alter the amount of HF on a violin.)

As before, an overall room pair should be added to complete the sound.

4.3.2 Microphone choice

The lowest fundamental frequency of the violin is G3, at 196 Hz, and so a
good quality condenser cardioid or omnidirectional microphone will be
sufficient to pick up the lower end and warmth of the tone. As with all
acoustic recording in a live space, the off-axis response is important when
collecting room reverb, but in the case of a solo performer, other off-axis
instruments do not need to be considered. Cardioids will pick up less of the
room reverb at the suggested height and would be a good choice in a very
live space. They will enable you to remain at a comfortable distance from
the instrument while keeping the amount of reverb under control.

Figure 4.4a

Figure 4.4a Change microphone angle to look over the instrument

In addition to the suggestions in Figure 4.4 for reducing the amount of

extreme HF picked up, ribbon microphones can also be considered, either
for the pair or as an additional central microphone (see Figure 4.5). The
smooth and gentle HF roll-off will ameliorate any harshness in the tone but
will still pick up some room reverb on the rear lobe. A useful technique for
a recording session, as previously noted, is to combine a pair of condenser
microphones mounted a little higher, at around 3 m (10′) (this increased
height will reduce the amount of HF slightly as it is absorbed by the air),
and a single ribbon microphone mounted about 30 cm (1′) lower.

Figure 4.4b

Figure 4.4b Lower the microphones

The ribbon can be placed closer to the violin because of its HF roll-off, and
this might be a useful asset where being closer is necessary, perhaps
because of the acoustic or some other constraint on microphone placement.
This combination of a pair of condensers plus a central ribbon microphone
allows you to use the ribbon microphone to fix the image more securely in
the centre if it exhibits a lot of lateral movement on the condenser pair
when the player swings around. The ribbon microphone will usually be
mixed in at about 6–8 dB lower than the condensers, so it plays a secondary
role in forming the sound, but it can be used to have a beneficial effect on
the tonality as well as on the image stability. It will give a warmth and
sweetness to the tone, while the condensers pick up more of the attack,
bowing noise, and other HF components. The violinist Itzhak Perlman
particularly likes ribbon microphones such as the RCA 44BX, and Coles
4038s were used while recording him in 1998 at the Saratoga Performing
Arts Centre in New York during his collaboration with the pianist Martha
Argerich.2

Figure 4.4c

Figure 4.4c Move them round in an arc maintaining height

4.4 Cello
The cello radiates from its body in a similar way to the violin, in that the
very highest frequencies are radiated perpendicular to the front surface and
the bridge, and the overall pattern is quite complex. However, in
comparison to the violin, it has been found that microphone placement is a
little less critical, and moving on- and off-axis from the bridge will affect
only the highest HF content. Its lowest fundamental is C3 at 130 Hz, which
suggests that a good cardioid will extend low enough in the frequency
range, although an omni might give a better tone.

Figure 4.5

Figure 4.5 High condenser pair plus a ribbon microphone on a

violin

The instrument’s playing position is very different to that of a violin in that

it rests on and is close to the floor. The reflections from the floor are
important to the sound, and so a solid wooden surface would be the
preferred choice when making a recording. A marble or stone floor is not
flattering because of its bright reflections, so if you have to record using
such a floor in a church for example, then put a small amount of carpet
down underneath the player’s chair. Carpet will absorb HF, so bear this in
mind when choosing to use it in other circumstances, and avoid using a
large area of it.

Some solo players like to sit on a podium to play because it can act as a
further resonator and enhance the sound, but a podium’s performance will
depend on how well it is constructed. Some will enhance the sound, but
others will absorb some of the LF, or produce an unpleasant resonance, or
simply be creaky and noisy as the player moves. If the podium is working
against you, you will have to negotiate with the player to find another
solution.

4.4.1 Microphone placement and choice

The technique outline for the violin can be adapted for the cello, using a
narrowly spaced pair of microphones (25–30 cm (10″ to 12″)) to give the
solo image more substance and width, and aiming to capture a good
balance of bowing detail, full tone, and sense of space. Because the playing
position is different to the violin, the microphones do not need to be as high
to capture the HF, but they can be placed in the region of 2.25 m (7′6″) high
and 1.8–2.1 m (6′ to 7′) away, depending on the room acoustic and the
microphones used. The microphones should be pointed towards the bridge,
perpendicular to the instrument’s front surface, and as you move the
microphones around in an arc L to R, you will find that the HF is
maximised when they are centrally placed in front of the instrument. Figure
4.6 illustrates the microphone position suggested as a starting point. In
order to get a good feeling of a real instrument in a real space, moving the
microphones a little closer or further away will help adjust the balance
between direct and indirect sound.

In a concert scenario, any microphone will have to be placed lower, but this
is less critical than it would be for the violin; there is a wider acceptable
range because the instrument is less directional in its radiation when floor
reflections are taken into account, and it is orientated differently with
respect to the microphone. A microphone low enough to be looking straight
at the bridge will be visually unobtrusive for live work, and a fig of 8
ribbon (Royer R-121 or Coles 4038) could be used to usefully discriminate
against the orchestra.

Condenser omnis or cardioids will work well, remembering that it is not

necessary to have a large-diaphragm microphone in order to capture a good
extended LF range. Any pencil omnidirectional microphone will have a
sufficiently extended frequency response.
The cello is also well suited to ribbon microphones using the HF roll-off to
produce a mellow sound, and it enables closer placement if necessary. At
the microphone distances suggested earlier, the proximity effect from the
ribbon microphone should not be at all evident, but it will begin to show at
the lowest frequencies at around 0.7–1 m (2′ to 3′4″) away – more likely in
a difficult live scenario. As with the guitar (which has a similar frequency
range), care should be taken not to make the instrument sound unnaturally
bass heavy and larger than life. Its tone needs both the warmth and the
more astringent upper frequencies to feel well balanced and real. Although
the ribbon microphones will miss some of the upper HF that would add
some bite, this part of the tone colour will be picked up on the overall room
pair of condensers.

Figure 4.6a

Figure 4.6a Microphone placement on cello – plan view showing

arc for HF adjustment

4.5 Woodwinds
In radiation terms, the woodwinds can be divided into the oboe, clarinet,
and bassoon, which all have reeds and are essentially closed at the
mouthpiece end and open at the bell; and the flute, which operates using a
stream of air striking the mouthpiece edge, and is open to the air at both the
mouthpiece and far end. This has an impact on the radiation patterns, and
thus which microphone placement options are most effective when
recording them.

4.5.1 Oboe, clarinet, and bassoon

For the first group of instruments, radiation comes from both the open
holes and the bell. The highest overtones pass right down the bore of the
instrument, radiating partly from the open tone holes but primarily from the
bell, forming a cone-shaped beam aligned with the axis of the instrument.
At the lower end of the frequency range (i.e. the fundamental frequencies
of the lower register), the radiation becomes stronger from the first open
holes than from the bell, and the overall radiation pattern becomes more
perpendicular to the instrument’s axis.3 The higher frequencies emerging
from the bell will be reflected from the floor (for the oboe and clarinet
whose bells point in that direction) and projected diagonally upwards from
the bassoon. The techniques used for the violin can be adapted for the
forward-facing woodwinds, although HF content will be best adjusted with
microphone height and angling rather than moving side to side as there is
no head-shadowing. You should avoid placing microphones directly on-
axis to the bell of the woodwind instruments, as you will collect more HF
and wind noise and miss out on obtaining a good overall balanced tone,
which is the goal of classical recording.

Figure 4.6b

Figure 4.6b Microphone placement on cello – side view

Microphones should be positioned in front of the player, but at a height of

around 2.25–2.6 m (7′6″ to 8′6″) and looking down (assuming the player is
standing; a bassoonist might prefer to sit and the microphones can be
lowered accordingly). Height is less critical for the clarinet and oboe than it
is for the violin because of the different radiation characteristics of the
instruments in conjunction with the floor reflections of the bell radiation,
although key noise needs to be avoided. As outlined previously, to obtain
an image that has some width and bloom around it, a pair of microphones
instead of a single spot microphone is used.

Figure 4.7

Figure 4.7 Generic end-blown woodwind radiation pattern at low,

mid, and high frequencies. This applies to overtones as well as
fundamental frequencies.

Key noise is the aspect of woodwind recording that will cause the most
difficulty, although bubbling and wheezing noises from worn or leaky reeds
can also be a nuisance, and it would be best to politely talk to the player
and see if they can change their reed where this sort of noise becomes
intrusive. Key noise is part of the instrumental sound, so it has to be
accepted to a certain extent, but it should not dominate. Keeping the
microphones at a good distance (above about 2.25 m (7′6″)) will help with
this, as the intrusiveness of transient-filled key noise is reduced by HF
absorption into the air. If the microphones are too close, all key noise, reed
noise, and mechanical sounds will be artificially exaggerated and feel very
unnatural to the listener, who does not usually hear the players close up. If
the room is reverberant, it will be better to avoid getting in too close (and
picking up too much key noise) by using directional microphones rather
than omnis. Ribbon microphones (Royer R-121 or Coles 4038) will also
ameliorate key noise by virtue of the HF roll-off inherent in the design.

4.5.2 Flute

The flute can be placed in a different category of radiation pattern because

it is supported more or less horizontally in its playing position, and its
radiation comes from both ends of the instrument as well as any open
holes.4 The result is that it radiates almost equally to the front and to the
back of the player, which can be very useful in opening up microphone
placement opportunities in less than ideal circumstances. Placing
microphones fairly close and halfway down its length should be avoided, as
some phase cancellations may result due to some frequencies radiating in
opposite phase from each end. The breath noise is more prominent from the
front, as it is formed by air turbulence as the breath strikes the top of the
mouthpiece. If too much breath noise is a problem, make sure the
microphones are not in front of the player or at the height of the player’s
mouth; because of the equal front and rear radiation, you will be able to
move the microphones behind the player to really reduce breath noise. Key
noise, as with all the winds, will be reduced with distance, so if you are
working in a reverberant space, keep the microphones at least a couple of
metres (6′8″) away from the instrument, but use cardioids instead of omnis
to reject a greater proportion of indirect sound.

Figure 4.8

Figure 4.8 Woodwind player with microphones – side view

Adding microphones from behind at approximately instrument height is a

good, discreet option in a live situation, and unlike for the violin, this will
not result in the loss of important HF content.

Figure 4.9

Figure 4.9 Flute – alternative microphone positions in front and

behind

4.6 Harpsichord
Although the harpsichord is clearly an early relation of the grand piano, the
technique of piano recording outlined in Chapter 5 cannot be simply
transferred over to the harpsichord; its radiation pattern is different, and
obtaining a balanced and characteristic sound needs a slightly different
approach.

4.6.1 Harpsichord characteristics

The instrument is not as loud as a modern piano; the string tensions are
lower, the soundboard technology is not as developed, and gradual changes
in dynamics are not possible as the instrument’s mechanism is not
responsive to the velocity with which the keys are struck. To facilitate
stepped changes in tone and dynamics, harpsichords have stops in a similar
way to organs, which are a means of coupling together sets of strings in a
variety of different octaves. Engaging and disengaging these will produce a
significant amount of noise, as the mechanisms are relatively simple and
primitive in design. The pedals can also make quite a lot of noise; placing a
small piece of carpet on the floor under the pedals can reduce this.

The largest instruments have a range of five octaves, from F1 at around 40

Hz to F6 at around 1400 Hz. The exact pitch is dependent on the tuning
being used; they are usually tuned to a lower pitch than A = 440 Hz, as
would be normal for instruments of the baroque era, whether they are
original instruments or reproductions. The upper frequency limit of the
whole spectrum is much higher than 1400 Hz because of the strong
presence of a multitude of higher partials. In order to fully capture the
lower notes, omnidirectional microphones would be preferred.
4.6.2 Microphone placement

The majority of the harpsichord’s useful sound is projected out to the side
away from the lid and towards the audience, and the effect of the lid
reflections is to add a layer of complexity and richness to the sound. When
recording the piano, the practice of approaching it from the tail end is in
part an attempt to reduce some of the complications that arise from the
muddying effects of the lid reflections. With the harpsichord, these same lid
reflections can be very useful in helping to integrate the sound and
effectively glue it all together. Increasing the lid’s angle so it is open more
widely will affect the openness of the sound, and this is something you
should experiment with. It is possible to remove the lid altogether, but the
effect will be to lose some of the characteristic complexity.

If microphones are placed around the tail end, the mid and lower string
resonances will be picked up, but not so much of the strong upper harmonic
components of the strings and the characteristic plucking sound of the
mechanism. All three are important aspects of the overall sound that needs
to be captured to give a balanced view of the instrument.

Figure 4.10 shows the harpsichord from above and the side, showing the
most useful arc in which microphones should be placed.

Given the lower overall level of output, it is usual to place the microphones
somewhat closer than for a grand piano, and a good place to start will be in
the region of 1.1–1.4 m (3′6″ to 4′6″) away from the well of the instrument
at a height of about 1.5–1.65 m (5′ to 5′6″).

Increasing the distance of the microphones away from the instrument will
reduce clarity and increase the amount of reverberation, but it will also
reduce the amount of mechanism noise, which is potentially useful.
However, the amount of mechanism noise can also be altered by the height
of the microphones without the need to move them further away if this has
a detrimental effect on the clarity of the sound. When the instrument lid is
fully open, a lower microphone position will pick up more of the plucking,
and a higher one will have more resonance and sustain. Microphone height
adjustment will also affect the amount and quality of lid reflections
collected, so it is worth spending some time altering this aspect of the
placement to find the best place on any given harpsichord. This is also a
good time to take a walk around the instrument and sit at different heights
to give yourself an idea of just how much the sound changes.

The overall aim is to find a happy medium where there is enough clarity (so
the recording is not dominated by room sound) whilst also ensuring that the
mechanism sounds are not unduly intrusive. Making sure that the sustained
part of the tone is well captured across the whole spectrum will help to
make listening to the recording really enjoyable. Listening to a harpsichord
recording where the transients are very harshly represented can become
fatiguing.

Figure 4.10

Figure 4.10 Harpsichord – plan view of pair of microphones

The microphones used in Figure 4.10 are a spaced pair at around 25–30 cm
(10″ to 12″), angled slightly outwards, in addition to an overall room pair.
The LCR three-microphone technique outlined in section 5.7.1 can also be
used on harpsichord, although they will be closer together as the physical
size of the instrument is not so great. Because of this, they could be panned
almost fully LCR, but the final width of the image should inform the
panning. In order to make the image and sense of space feel real, this
technique will need more support from some sort of overall room pickup to
act as a unifying element.

Notes
1Fundamentals of Musical Acoustics, sections 24.5
Arthur H. Benade
Publishers: Dover
ISBN 0–486–26484-X
2MUSIC FROM SARATOGA/BEETHOVEN/FRANCK Itzhak
Perlman/Martha Argerich/EMI CLASSICS (1999) UPC 717794908725
3The Physics of Musical Instruments, section 15.5
Neville H. Fletcher and Thomas D. Rossing
Publishers: Springer
ISBN 978-0-387-98374-5
4The Physics of Musical Instruments, sections 16.11.2, 17.7
Neville H. Fletcher and Thomas D. Rossing
Publishers: Springer
ISBN 978-0-387-98374-5