Those are the words which Thomas Edison
himself first recorded on the phonograph he invented in 1877.
At the time, his tinfoil records could be played back only a couple of times
before the stylus ruined the groove.
(2017-11-03) Room Acoustics. Dry room, audio booth, sound stage.
Avoiding two pitfalls: Outside noise and inside echoes.
You can listen to recorded sound anywhere and you can record it anywhere.
In either case, however, unacceptable degradation will result
if a few simple precautions are not taken concerning isolation from
outside noise and prevention from echoing within the room.
(2018-02-03) Microphone Stands and Booms
By far, the most common thread for microphone mounts is 5/8''-27.
5/8''-27 means a diameter of 5/8'' (15.9 mm)
and 27 threads per inch.
In the old days, the microphones themselves were threaded.
Such was the case for the legendary Model 55 Unidyne Microphone
introduced by Shure Brothers
in 1939 and once described as the most recognized microphone in the world.
The so-called Elvis microphone was a scaled-down version of the 1939 model introduced in 1955.
This vintage look is still preserved by Shure in current "55" microphones
which encase different shock-mounted modern capsules (Super 55 and 55SH II).
With that notable exception, today's full-sized microphones are rarely threaded, if ever.
Instead, they fit into threaded mic clips
or shockmounts, which provide an external isolation from vibrations.
Because microphones are so often used in close proximity to cameras,
they sometimes have to share the same threaded studs and it's useful to keep
a couple of
handy, to fit either 3/8''-16 or 1/4''-20.
(2018-02-07) Pop filters = Pop screens = Pop shields
Eliminating plosives and guarding from spit.
For studio mics at least, I like
cylindrical pop filters best
($10 a piece).
They're even more effective (with directional mics) if you
point the microphone at your mouth, but not your mouth at the microphone.
This way, you can look directly at a camera without hiding your face behind the
microphone (again, cylindrical pop filters are the least obtrusive ones in that situation too).
When designed by people who know their craft, all computer graphics
look as if they're lit from the upper left of the screen.
(This applies, in particular, to common action buttons.)
If at all possible, a video meant to be viewed primarily on busy computer screens
should be lit the same way (key light to the left of the camera or to the
right of the talent).
Consequently, when a studio microphone is visible at the same level as the face of
the talent, it's often best to put it to the left
of the talent so the key light won't cast a shadow from the microphone on the face.
(2018-01-14) Volume = Perceived Loudness
Loudness normalization to broadcast standards.
Humans may perceive the loudness of identical
sound levels (dB SPL) differently according to their
For normalization purposes, the most commonly used calibration of that effect
(in North America, at least) is called A-weighing.
Decibels of perceived loudness, following that standard ponderation, are
indicated by the symbol dBA. This can be used to measure the loudness of various
types of sounds containing diverse mixtures of frequencies:
Ear damage is likely after prolonged exposure (8 hours).
Propeller-plane flyover (1000 ft)
Vacuum cleaner (3 ft)
Passing diesel truck
Busy New-York street.
Vacuum cleaner (10 ft)
Quiet suburb (daytime)
Urban living room
Quiet rural area
Desert without wind
Conventional limit of audibility
If you double the distance to a localized sound source, you reduce its loudness by 6.02 dB
(20 log 2).
Tripling that distance reduces the loudness by almost 10 dB (9.54 dB).
Multiplying the distance to the sound source by 10 reduces perceived loudness by
exactly 20 dB.
A 1 kHz sound at 1 dB SPL is nearly 1 dBA, by definition (IEC 651).
(2018-01-15) Automatic Gain Control, Limiters and Compressors
Three ways to make the best of imperfect recording setups.
The input levels of an audio recording device are best adjusted manually to
account for the different sensibilities of
microphones and widely varying recording conditions (intrinsic loudness and
distance from the sources).
(2018-01-25) Rating the Sensitivities of Microphones
Sensitivity is the ratio of voltage output to sound pressure input.
The sensitivity of a microphone is defined as the ratio of the
variation in its electromotive force output
(i.e., the open-circuit voltage it produces) to the corresponding
variation in sound pressure input.
When neither is too large, that's a constant.
That sensitivity is most commonly expressed in mV/Pa
(millivolt per pascal) assuming a standard sinusoidal sound signal of 1 Pascal amplitude (94 dB SPL)
at 1 kHz.
When decibels are used,
the dBV/Pa scale is usually understood.
Thus, the stated dB sensitivity rating is 20 times
the decimal logarithm
of the sensitivity expressed in V/Pa
(which is the standard SI unit for sensitivity).
Sensitivity of 35.5 mV/Pa is -29 dB (re: 1 V at 1 Pa).
Indeed, we have: 20 log (0.0355 V/Pa) = -28.99543... dBV/Pa
A microphone is said to be loud when its sensitivity is high
(the opposite of loud is soft).
Normally, the louder the better, because of a lesser need for amplification
(as a signal is amplified, so is the accompanying noise).
However, there's such a thing as too much of a good thing: If a microphone is too loud,
it may generate unexpectedly high voltages at the input of the next circuits
That won't damage them but they'll saturate or clip
(same thing) which will introduce unacceptable distortion.
Today, microphones are rarely designed with a sensitivity exceeding 50 mV/Pa
(that's -26 dBV/Pa). The discontinued predecessor of the
aforementioned BP4071 was the AT4071a which had a nominal sensitivity of
89.1 mV/Pa (that was -21 dB). This is considered
way too loud by today's standards for general-purpose use
(although bird lovers still like this kind of sensitivity in a directional microphone).
The sensitivity rating of a microphone is always stated for on-axis sound
(coming from the preferred direction of the microphone, which gives the maximum sensitivity).
Polar patterns are charts giving relative sensitivity as a function of direction
by reference to that basic sensitivity
(de facto, the 0 dB level for a particular microphone).
An older decibel scale for microphone sensitivity is still floating around which differs from the modern
one by 20 db. In that obsolete scale, the aforementioned BP4071
would have been quoted as having a sensitivity of -49 dB (which could be misinterpreted as quite low).
The discrepancy comes from the former use in acoustics of the units of pressure still preferred by
many meteorologists. When meteorologists in France and elsewhere were criticized for issuing TV reports
expressed in millibars (1 mb = 100 Pa) instead of
a proper SI unit, they didn't change their numbers
but started using the SI-equivalent hectopascal
(which is correct albeit arguably somewhat weird).
In the old days, audio engineers were routinely using the microbar (0.1 Pa)
as their unit of pressure (this was an alternate name for the official unit of pressure
in the CGS system, namely the dyne per square centimeter).
This made the volt per microbar (10 V/Pa)
their unit for microphone sensitivity. Using that unit, the
numerical values of sensitivities are ten times smaller than the values in V/Pa.
Expressed in decibels, they're thus 20 dB lower, as advertised!
In Numericana, microphone sensitivities are only given in mV/Pa.
Not only does this avoid the aforementioned ambiguity of decibels,
but it also serves as a constant reminder of what sensitivity is all about.
Unit conversions is a known source of distress.
NASA once crashed a spacecraft on Mars because of that.
As I was looking for data relevant to this page,
I came across a
discussion between audio afficionados
where the above 20 dB offset is mistaken for real substance.
Electromotive force vs. measured voltage :
Microphone sensitivities are best expressed in terms of open-circuit voltages
(the common name given to electromotive forces).
So defined, sensitivity depends only on the microphone itself,
not on whatever load it may have to drive (usually, the input impedance A of some preamplifier).
Nevertheless, some manufacturers give the actual voltage V that would be observed
per unit of sound pressure into a load of specified impedance within the range they recommend
(e.g., A = 1 kW).
That number is slightly smaller than the aforementioned intrinsic sensitivity U.
The relation between the two is obtained by observing that the microphone's output current is equal to
the amplifier's input current.
Let R be the microphone's output impedance and A be the amplifier's
input impedance. This means:
U / (R+A) = V / A Therefore:
U = ( 1 + R/A ) V
For example, if a 200 W microphone is said to yield
10 mV/Pa into a 1 kW load,
then we just quote its intrinsic sensitivity U as 12 mV/Pa.
Such a microphone would drive a 2.5 k load with
12 / (1 + 200 /2500) = 11.111 mV/Pa.
(2018-02-05) Microphone directionality. Polar pickup patterns.
Omnidirectional, bidirectional, cardioid, etc.
(2018-02-05) Dual-Diaphragm Microphones for variable pickup patterns.
A cardioid is the sum of omnidirectional and bidirectional patterns.
Multi-pattern and variable pickup pattern:
Mixing equal parts of an omnidirectional pickup pattern and a bidirectional one
(figure-8) yields a cardioid pattern.
Other proportions in this type of mixing also yields the other traditional
microphone pickup patterns:
Subcardioid pattern (between omnidirectional and cardioid).
Supercardioid pattern (between cardioid and figure-of-eight).
The above five patterns and the four intermediate ones between them yield a palette of nine patterns, which
are commonly available at the flip of a switch in somemulti-pattern microphones.
The term hypercardioid is sometimes applied to any such mix but it's most
often reserved to patterns with a wide cone of silence up to
the bidirectional figure-of-eight itself (90° angle of silence).
Variable-pattern microphones (e.g., CAD Audio M179) even allow you
the freedom to dial anything in-between.
Note that a two-diaphragm microphone which is capable of recording at least two
pickup patterns from the above standard family can also be used to reconstruct
any of them in post-production. For example, if the omnidirectional signal is on the
left channel and the figure-8 bidirectional signal is on the right channel,
Then, you obtain a forward cardioid by adding the two channels and a backward cardioid by subtracting
them (opamps were originally intended do perform exactly this
sort of operations).
This way, you can choose the pickup pattern after recording.
Any dual-diaphragm microphone could be modified into a pseudo-stereo microphone capable
of recording the two separate phase-tracks just described.
Two products offer this capability straight out-of-the-box:
The Lewitt LCT 640 TS microphone (TS stands for twin system)
whose secondary output is on a mini-XLR connector (sideways).
("Mid Side") attachment for the 10-pin connector of Zoom recorders
(H5, H6, F1, F4, F8)
records separately the signals from two capsules: Forward cardioid and
Other multiple-diaphragm configurations would provide other capabilities.
Multi-pattern dual-diaphragm microphones (data is for cardioid pattern)
Most prosumer measuring microphones have a ¼'' capsule (6 mm diaphragm).
Professionals sometimes use smaller membranes (3 mm)
which are more accurate in the upper-part of the audio spectrum.
They tend to prefer expensive low-noise ½'' units for general use.
Acoustic Calibrators (1 kHz, 94 dB SPL) :
By convention, absolute calibration of a sound-measuring instrument
is always done at 1000 Hz. For that purpose, standard sound sources
are available which deliver precisely
94 dB SPL into a force-fit
microphone port allowing cylindrical microphone heads up to 1'' in diameter
(sometimes only ½'').
Smaller microphones require adapters which may or may not be included with calibrating units.
Below is a list of current models of such acoustical calibrators.
All of these can work either at 94 db
or 114 dB (the latter setting is helpful in a noisy environment).
Now, the calibrators themselves drift out of calibration and have to be recalibrated yearly
by the manufacturer. Most people will only trust Brüel &
Kjær (or possibly Cirrus) for that follow-up.
Sound Meters, Measurement Microphones :
Measurement microphones are designed to be as linear as possible,
They have a flat frequency response throughout the audio range and
deviations must be carefully documented (see example below).
Tiny diaphragms help keep resonant frequencies safely outside of the audio domain.
Many uncalibrated consumer models are just intended for the analysis of room acoustics
and cannot be trusted beyond a precision of 2 dB or 3 dB.
The following models are thus not recommended for scientific applications:
A much better precision is offered at a similar cost with any of the models listed below.
Each such unit comes with an individual calibration curve made with a professional instrument.
The resulting on-axis frequency response is typically made available
online (tied to the serial number of every microphone)
in a digital form suitable for audio-analysis software.
Sonarworks also provides an off-axis curve.
Some measurement microphones which come with individual calibration curves :
With a street price of $50 (I got mine on sale for $40) the EMM6 is the most
affordable of the above. Packed with each unit is a dated plot of its frequency response.
The corresponding data is also available online (tied to the serial number)
in the form of a tab-separated text file
(ready to import into Excel or other specialized software). That file contains
measurements at a precision of 0.1 dB for 256 frequencies
whose logarithms are evenly spaced, from
20 Hz (n = 0) to 20000 Hz (n = 255). That's to say:
fn = (20 Hz) 10 n/85 (for n between 0 and 255 = 3 x 85)
The values are given in decibels relative to the level at frequency f145 = 1016.0436 Hz
which is given tersely in absolute terms (in dBV/Pa) on the first line of the data which reads,
in the example of my own unit:
This misleading header actually indicates that the sensitivity of this particular microphone is
-39.6 dBV/Pa (i.e., about 10.47 mV/Pa) at precisely 1016.04 Hz (not 1000 Hz).
The blow-up at right shows the frequency-response near 1 kHz of my own new EMM6.
Each black square is precisely a single data-point (it covers exactly one pixel in the
full graph shown below, where the height of each pixel is just 0.1 dBV/Pa.)
To obtain a very precise value of the sensitivity at exactly 1 kHz (which is the usual standard)
we remark that 1000 Hz = fn when
The data for my own unit says that the sensitivity for
f144 is 0.2 dB above the level for the aforementioned ad hoc
reference frequency (f145 ).
Thus, the response at 1000 Hz is best obtained by linear interpolation:
(2018-02-09) Low Cut Filter = High Pass Filter (HPF)
Filtering out the lowest audio frequencies.
On many microphones, a switchable low cut filter is provided
to get rid of the low-audio and sub-audio hum and rumble.
Typically, a corner frequency of 80 Hz is used.
Most manufacturers are content with a simple
first-order filter (6 dB/octave) which
provides a modest 12 dB attenuation at 20 Hz.
Others, like Audio-Technica
will do more and they should be commended for it.
If you need low-cut in an urban environment, the more attenuation the better.
Even in their entry-level AT2035, the low-cut filter
they provide is second-order (12 dB/octave)
for a 24 dB attenuation at 20 Hz.
On Audio-Technica shotgun microphones
(AT897, BP4073, BP4071) the switchable low-cut filter is third-order
(18 dB per octave) and provides an attenuation of 36 dB at 20 Hz
(that's 12 dB at 50 Hz).
(2018-02-09) Low Pass = High Cut
A good low-pass analog filter is paramount for proper digitization.
Everything above 20 kHz is utterly useless.
The human ear is unable to detect is. Only badies can hear 20 kHz.
Young adults are lucky if they can detect a sinewave at 18 kHz.
Middle-aged people can hear a thing above 14 kHz or 15 kHz, at best.
Furthermore, with a 48 kHz
sampling rate anything above 24 kHz will actually damage
the digitized audio signal beyond repair (in the form of additional audible noise).
For the utmost in quality, we must attenuate everything in that part the spectrum
as much as possible with analog filteringprior to digital sampling.
Any leftover ultrasonic component (beyond exactly 24 kHz)
will cause audible noise in the digitized audio signal.
Do keep that in mind if you happen to use a fancy
Earthworks microphone with an unusually wide bandwidth.
Those need more low-pass filtering...
In the analog to digital conversion process,
any ultrasound translate into muddy hiss, not added clarity!
(2017-11-22) Characteristics of Full-Size Wired Microphones
Condenser type (varying capacitor) or dynamic type (varying inductor).
Microphones currently being produced range in price from $1.67 to
thousands of dollars
(the AKG C12 VR sells for &5999).
Used vintage Neuman U67 tube
LDC microphones are typically sold for $9000-$16000, depending on condition.
At least part of that madness is due to a nostalgia for the particular type
of distortion introduced by
tube (or valve) circuits.
For some obscure reason, tube amplifiers tend to distort a waveform symmetrically,
which is another way to say that they introduce more evenharmonics than odd ones.
The type of sound so produced is normally associated with female voices
(it would seem that Adam's apple
on a male voice box
is responsible for producing asymmetrical waveforms rich in odd harmonics).
Whatever the exact reason may be, this is just one example of an acquired taste
among audiophiles which has little or nothing to do with high-fidelity.
If anything, modern semiconductor circuits have better fidelity qualities,
The good news is that those high-price instruments are not a necessary part
of high-fidelity home recording.
Designing microphones is an art form in itself.
Microphones are a crucial tool for musicians and an object of worship for countless
Just enumerating the main aspects on which that subculture is based will serve to
demonstrate that we can only scratch the surface here
(focusing, as usual, on nontrivial numerical aspects besides cost).
All these aspects are interrelated:
Price, cost of ownership.
Options and customizability.
Look, feel and durability.
Size and weight.
Possible mounts (handheld, tabletop, lapel, stand, arm, boom).
Sensitivity at various frequencies (bandwidth & microstructure).
Impedance magnitude and phase shift (as functions of frequency).
Directivity (polar pattern) at various frequencies.
Proximity effect at various frequencies.
Noise figure, noise floor (hiss).
The previously introduced concept of sensitivity
influences greatly overall noise performance because lower sensitivity
demands greater subsequent amplification, which magnifies hiss just as much as the useful signal.
The self-noise (or equivalent noise level,
henceforth tabulated as hiss)
of a microphone is the loudness of the signal it produces by itself in an isolated soundproof enclosure
(it would be cheating to report only the electric noise of the apparatus without the microphone capsule).
The same figure of merit is sometimes reported as a
signal-to-noise ratio (SNR) assuming a 1 kHz
sinusoidal standard soundwave of
1 Pa amplitude (94 dB SPL):
SNR = 94 dB - (self-noise, dB)
The dynamic range of a microphone is defined as
the decibel difference between the aforementioned self-noise
and the top loudness it can record, with less than 1% THD
(total harmonic distortion).
The nominal output impedance is expressed in ohms (W).
A microphone is normally plugged into a
preamplifier whose input impedance shouldn't be lower than whatever
is specified by the microphone manufacturer
On the other hand, it shouldn't be too high either because high impedance breeds noise.
A time-honored rule of thumb is to load a microphone with five to ten times its own output impedance.
Some compact microphones sold with on-camera mounts :
(2018-02-01) Acoustical Properties of Large Circular Diaphragms
Resonant frequencies and frequency-dependence of pickup patterns.
The diaphragm of a condenser microphone consists of a thin circular membrane whose rim is
attached under tension to a rigid hollow cylinder.
In the so-called center terminated variant,
the diaphragm is also anchored by a small screw at the center,
where it can neither move nor tilt...
That method is used, in particular, in good ½''measurement microphones.
It presents three major advantages:
The center point can be used for electrical contact.
Resonances are suppressed if the center isn't a node.
Resonances are suppressed if the gradient at the center is nonzero.
Those last two properties eliminate the lowest resonant frequencies for a
circular membrane of prescribed size, areal weight and tension.
That helps remove all resonant frequencies away from the audio range.
However, the central contact restricts the amplitude of the diaphragm's motion at lower frequencies
and thus reduces the basic sensitivity of the microphone.
A condenser microphone is formed by the varying capacitor consisting of one such diaphragm
opposite a rigid backplate (polarized by an external voltage and/or an
When those two form a closed capsule, an omnidirectional
pickup pattern is obtained (at least at low frequencies).
A microphone capsule is never completely closed,
or else it could bend (or even pop) in response to slow changes
in atmospheric pressure. There are just tiny vents which allow air
to go in and out of the capsule fairly slowly, with little or no impact at audio frequencies.
The best designs will make the vents just large enough to cancel
hum just below the audio range (which is usually assumed to start at 20 Hz,
although that's definitely not audible).
The mathematical simplicity of the above configurations makes a complete
theoretical analysis possible, which may serve as a useful basis for experimental refinements
in the actual design of commercial microphones.
Another aspect amenable to pencil-and-paper analysis (barely so) is
the pickup pattern (sensitivity as a function of direction)
of a large-diaphragm for a sound having a wavelength commensurate with its size
(for much larger wavelengths, the pickup pattern is omnidirectional).
(2018-01-22) Large-Diaphragm Condenser Microphones (LDC)
Quintessential capacitive microphones. Every voiceover artist has one.
Most condenser microphones use the 48 V phantom power normally
found on XLR sockets (one more reason to get an
XLR1 audio adapter, if you shoot video with a Panasonic Lumix GH5).
What's the fuss about large diaphragms?
Well, the larger the diaphragm the quieter the microphone,
but too large a diaphragm will struggle with the upper part of the audio spectrum,
especially off-axis, as different parts of the membrane see different phases
of the soundwave (at 20 kHz the wavelength is only 17 mm).
Earthworks achieved the extreme bandwidth (30-33kHz) of their
SV33 flagship by limiting the diameter of the diaphragm to 14 mm.
The price they paid was a 15 dB noise-floor which is unimpressive
for a microphone at that price-point ($2499).
The membranes of more typical LDC microphones are about 1''
in diameter (25.4 mm).
One popular LDC microphone is the affordable
AT2020 from Audio-Technica
I went instead for its big brother, the
because of a side-by-side sound comparison on YouTube.
Also, unlike the AT2020, the AT2035 has two desirable features:
Switchable 10 dB in-line attenuator ("pad")
whose effect is equivalent to tripling the distance from the sound source.
Switchable second-order high-pass filter
with 80 Hz corner frequency, which helps cut out hum and rumble in an urban environment.
(Other makes often provide only first-order.)
The AT2035 gets
as the best in its class (I wouldn't consider a higher class for home use,
following the law of diminishing returns).
That microphone comes with a soft pouch and a shockmount (including
a plastic thread adapter; 5/8''-27 male to 3/8''-16 female).
I got mine with a complimentary 10-ft XLR cable and Neewer® pop screen.
All for $149.
The shockmount by itself (AT8458) would sell for
(Third-party shockmounts go for
a short cable is about $9 and the pop shield is $7.)
The AT2035 was released in 2008.
It's built around a center-terminated 24.3 mm diaphragm (0.96").
It uses back electret polarization,
which helps accommodate a wide range of phantom voltages (from 11V to 52V).
Some purists still scoff at this approach, compared to what they call true
condenser microphones, in spite of the fact that the electret technology has been around
for more than 50 years and helps deliver superb performance.
To address such queasies within the Audio-Technica ladder, the AT2035
is bracketed by two condenser microphones which are purely DC-biased without electrets,
the AT2020 and the AT4040 (which both demand 48V phantom power).
The latter costs twice as much as the AT2035 without offering any improvement in self-noise.
(Since it's 1 dB more sensitive, it's technically just 1 dB quieter.)
This isn't the whole story, though:
The noise figure of the AT4040 was achieved in spite of the fact that it uses only
a smaller diaphragm of 20.4 mm (0.8'')
which helps with transient response.
Although my own ears couldn't detect those subtleties
(I'm now on the wrong side of sixty) I could easily see that the AT4040 grille is more transparent,
which can be acoustically desirable.
Røde's NT1-A still looks like a better upgrade,
as a true condenser microphone which is 8 dB quieter
than the AT2035 (for only $80 more).
For another $40, I find their NT1 even more tempting with its true-to-life
flat-response sound and praised shockmount (Rycote lyre).
reports that they incorporated into the AT2035 the honeycomb diaphragm design used in their own
flagship, for increased surface area and enhanced performance.
In the following table, we give a wide selection of the medium-to-large condenser microphones
available today. All of those are single-diaphragm microphones
(we list separately dual diaphragm microphones featuring selectable
They're all cardioid microphones, except :
Earthworks SR40V (hypercardioid).
CAD Audio E100s, (supercardioid).
Some LDC microphones (Data with all pads and filters disengaged.)
Because of its 4 dB sensitivity advantage,
Audio-Technica's AT2035 ends up being 12 dB less noisy than the AT2020
(or 18 dB less noisy than the multi-pattern Behringer C-3).
Likewise at the high-end, the AT5040 is 8 dB more sensitive and 15 dB
less noisy than the AT2035.
It's twice as sensitive and 4.7 dB less noisy than the Equitek E100s.
The Samson C01 mic gets mixed reviews; it's reportedly rather hissy.
(2017-11-22) Dynamic Microphones (French: bobine mobile)
Rugged inductive microphones, usually with limited bandwidth.
A moving-coil dynamic microphone functions exactly as an ordinary speaker. Actually,
a moving-coil speaker can be wired to work as a dynamic microphone, albeit a lousy one.
Because a dynamic microphone is a passive component, it generates no noise besides thermal
Unlike condenser microphones, dynamic microphones don't require
any outside polarization voltage to work.
There are two very different types of dynamic microphones:
As part of an old-school PA system I purchased years ago, I got the rugged
Radio-Shack 3303043 Super-cardioid Dynamic Microphone (RS catalog number 33-3043)
which is a perfect voice microphone in that capacity (great proximity effect).
That unit is still available new on eBay, between $25 and $50 or so
(it goes for less than $20 used).
It has the exact same look and feel as the legendaryShure SM58S
(SM58 with a mute switch). Both feature the exact same spherical grille (51 mm diameter).
The built quality is the same, except that the RadioShack body is a half-inch longer and has a black grille
coupling (which is silver on the Shure unit).
Some Dynamic Microphones (moving-coil microphones)
The 33-3043 microphone was manufactured by Shure specifically for RadioShack.
So were other dynamic microphones. All were made in Mexico and none
had any direct equivalent in the regular Shure line (they were typically loosely
related to more expensive Shure models sporting the same grille). Examples include:
(2018-02-04) Ribbon Microphones
A very special type of dynamic microphone.
The engine (or motor) of a ribbon microphone
is a very thin corrugated strip of metal (usually aluminum)
which fits tightly between very strong magnets without touching them
(today, neodymium magnets are used). The ribbon thus separates two
symmetrical cavities formed by the walls of the magnet.
As the ribbon moves in response to sound pressure,
a tiny electromotive force appears between its extremities which are
connected to the primary windings of a step-up audio transformer.
The natural acoustical symmetry of such microphones translate into a figure-8
They pick up sound equally well from the front or the back and very little from the
Ribbon microphones include legendary lip microphones like the
Coles 4104 for dramatic voice reporting in very loud environments.
Examples of Ribbon Microphones (special type of dynamic microphones)
(2018-02-02) Shotgun Microphones
Small-diaphragm condenser microphones with high directivity.
A shotgun microphone consists of a standard standar capsule monted at the rear
of a long interference tube with a number of slots on it.
On-axis sound passes through the tube unimpeded or theough the different slots in phase
(constructive interference). On the other hand,
destructive interference attenuates off-axis waves as they pass through the slots with
Because of their natural cylindrical shape, shotgun microphones often
feature a compartment for a single AA battery to power them as
an alternative to phantom power
(units primarily intended for use with DSLR or
hybrid cameras don't even allow phantom power).
According to the specifications of Røde and
other manufacturers, the battery must be
a 1.5V cell
(i.e., a single-use alkaline battery).
A rechargeable NiMH battery has a nominal voltage of only 1.2V,
which makes it unsuitable.
A well-conditoned fully-charged NiMH cell may work at first
(the initial voltage of a freshly-charged battery is about 1.46V) but it will
struggle and fail very soon. You've been warned.
The Audio-Technica models AT4071a and AT4073a are discontinued.
They've been superseded by the BP4071 and BP4073, respectively.
A few comments are needed about the bottom of that table, which lists low-end consumer product,
as the listed prices indicate:
The VidPro models (14-inch XM-88 and 10-inch XM-55)
come with plenty of accessories (each as a 13-piece kit in a molded case).
Their noise figures are undisclosed by the distributor. The audio quality is modest but
either microphone can be very cost-effective, as it can be plugged directly into the 3.5 mm socket
of a DSLR (cable included) running off its own internal AA battery. They can also use XLR phantom power.
The BY-PVM1000 is consistently reported to suffer from crackling noises when operated off 48V phantom power.
This problem is reported in some written reviews and can be heard even in
favorable video reviews.
That seems to be a design flaw present in all units
(it may be caused by capacitors with borderline voltage ratings).
Not recommended at all for use with 48V phantom power (and audio quality is downrated on battery power).
Could be OK with 24V phantom power, who knows?
The cheapest XLR shotgun microphone, sold as Marantz SG-5B, is just adequate for experimentations
and educational projects (dissecting a microphone).
It has been on sale at $16 or less.
Its restricted bandwitdth and high noise make it unsuitable for any type of video production.
(It's apparently not a fake; the official Marantz site does report the poor specs.)
For completenes, the Neewer bargain brand also sells short (10")
and long (14.37") shotgun microphones on the cheap
(for $23 and
respectively). They can't use phantom power and will work for up to 26 hours
off a single AA battery.
(2017-11-01) Lavalier Microphones (Lapel Mics) :
The best way to isolate a voice from ambient sound.
It's an unavoidable part of the physics of sound
that tiny microphones will produce more hiss than full-sized ones.
Lavalier mics are appealing in other ways. Draw your own conclusions.
All commercially available lapel mics are condenser mics which
need either their own battery or plug-in power
from the audio socket, typically from 2 V to 10 V
(more than 10 V may damage the mic and 48 V will fry it).
Properly taking sensitivities into account, the shocking truth which emerges from
the nonexhaustive table below is that the least noisy lavalier mics are the ME2 and the Giant Squid (the latter
being only 0.2 dB behind, which isn't significant).
The MKE2, which costs three times more than the former and eight times more than the latter,
is actually 2 dB worse than either!
The J 044 and the HQ-S are respectively 5 dB
and 10 dB worse than the ME2.
(I don't have data yet for the Purple Panda and the lowly Neewer.)
Noise is only part of the whole story and the less-than-stellar performance of the
expensive MKE2 in that department is entirely due to its tiny size.
The relatively low noise of the ME2 is partly due to its limited bandwidth.
Some Omni-Directional Lavalier Microphones (a.k.a. Lav mics, lapel mics)
Sennheiser's cost-no-object MKE2 is fairly bright (+4 dB at 10kHz) to compensate for the
fact that it's normally worn under a shirt. It comes with several caps to adjust its frequency response.
Sennheiser's mics come with locking plugs ("EW" = "Evolution Wireless").
JK's very popular Mic-J 044 (which may well be the best value for the money)
is available with many plugs to choose from (including Sennheiser's locking connector).
Usually, all others only have regular TRS and/or TRRS 3.5mm audio jacks.
The Neewer 0077 microphones are extremely cheap
(I just got three of them for a grand total of $4.99.)
You can't buy fewer than three at a time. They are essentially disposable microphones.
They are reportedly prone to failure
and are supposed to produce only junk boomy sound... However, they're certainly
not a total waste of money. They do sound better than
most on-camera mics. With low expectations, I was even
surprised to find the sound rather pleasant on my initial test!
The actual street price is $400 but the bundled lapel mic retails for $250...
Røde also sells the NewsShooter kit
with the same receiver paired to a more flexible transmitter, featuring both an XLR
socket (with phantom power) and a 3.5 mm socket
for third-party lavalier mics.
The above price is only an estimate
of what the system would cost if it was sold without a
lavalier mic, which isn't the case
(it's actually either bundled with a $150 ME2 for $700 or with a $400 MKE2 for $900).
The system uses proprietary rechargeable batteries and, arguably, you should have
at least one extra battery for the transmitter and one for the receiver,
for an additional cost of $100 or so (there are no third-party suppliers).
Sennheiser's AVX system is the digital successor to their very popular EW100 G3 analog model.
It's the Rolls-Royce of wireless microphones;
superb user-friendly engineering at a hefty price. Extremely easy to use.
What made me buy it in spite of the cost is the small size of the receiver,
which is a perfect match for the XLR1 (whose other
XLR mic input can then be used to capture ambient sound on the other audio track).
The radio part is designed from the ground up to provide no interference from any source
in the foreseeable future.
Both units are constantly in two-way communication to maintain a clear channel
in the allotted band. All data is continuously transmitted on two separate channels
for seamless switching from one channel to the next if needed.
Up to 8 AVX systems can share the same airspace and negotiate between themselves
for trouble-free communications without any human intervention.
For good measure, the audio data is encrypted, to prevent electromagnetic eavesdropping.
Radio communications are entirely digital, using GFSK modulation
(Gaussian Frequency Shift Keying) which is to say that the digital signal
passes through a Gaussian filter before being frequency modulated
(this method allows narrower radio bandwidth; it's what Bluetooth® uses).
The audio signal is digitized with 24-bit resolution
at a 48 kHz sampling rate.
We're very far from yesteryear's one-way analog transmission of an audio signal
over a single analog FM channel selected once and for all among a dozen choices or so.
In Sennheiser's parlance, the small receiver (actually a transceiver)
is called EKP AVX and its battery is BA 20.
The body pack (SK AVX ) takes a BA 30 battery.
You need at least one extra BA 30 for prolonged use,
since the body pack cannot be recharged while in use (it's on a untethered
moving body, after all). Either battery can be recharged whether it's
mounted to the corresponding unit or not, using a standard USB-A to
from any powered type-A socket (one such cable and a small AC adapter are included).
The LED indicator (red when charging, green when fully charged)
isn't designed for color-blind people.
Also available is a handheld microphone (SKM AVX) which takes a third
kind of battery pack (BA 10) which, surprisingly,
can't be charged when mounted (unlike the other AVX batteries).
It takes 4½ hours
to fully charge a BA 10 or BA 30
(good for up to 15 hours of continuous use).
The tiny BA 20 battery can be fully charged in 1¼ h
and will power the EKP AVX receiver for 4 hours.
A 4-level battery status is provided when the left button is pressed.
A blinking alert indicates there's less than 15 minutes of battery power left.
You can power the EKP AVX with the USB cable for an unlimited time
when working tethered.
If an EKP AVX receiver is plugged into an XLR socket with
phantom power, it will turn on and off automatically (to save battery power)
by sensing the presence of power in the socket. One less switch to worry about.
The EKP AVX turns itself off about 10 seconds after it
sees phantom voltage drop. Modern cameras take some time to switch off
and there may also be a significant delay due to the slow discharge of capacitors
with little resistive load on them. All told,
an EKP AVX connected to an XLR1
(with phantom power) on a Lumix GH5
turns itself off about 13 seconds after the camera is switched off.
One benefit is that there's no loss of pairing if you reset the camera
by power-cycling it for whatever reason.
On the other hand, be aware that the GH5 turns the XLR1 off
(along with phantom power) when it's used for previewing clips
on the back of the camera. To get out of this power-saving mode,
half-press the shutter button and wait up to 10 seconds for a new
pairing to take place (the camera shouldn't be too far from the mic).
The SK AVX bodypack input socket accepts either a line input or a microphone
(including third-party replacement microphone with Sennheiser/Sony locking jacks).
The bandwidth for a line source is 20Hz to 20kHz. For a microphone, it's
limited to 50Hz to 20kHz (which is more than enough).
EW plug (Evolution Wireless) is a 3.5 mm locking audio plug with
It's used for audio input from either a microphones (tip) or line signals
(ring; 1 MW input impedance).
Whichever input is unused must be grounded (i.e., connected to the shielding sleeve)
within the input plug and/or the input device.
The AVX system has a constant latency of 19 ms
which would correspond to a sound source located 6.5 m (21 ft)
away from the listener. There's usually no need to adjust that in post-production,
except possibly for an extreme close-up shot of a person talking, in which case
reducing the latency ought to be reduced down to 1 ms or 2 ms
(never less) to reproduce more precisely the time-delay our brains are accustomed to
when carrying a conversation up close (1 ft or 2 ft away).
most accustomed to.
Likewise, if you have to synchronize sound based on the image of a
clapper, make sure you introduce a delay corresponding
to a delay of about one millisecond per foot of distance between the camera and the subject
(the brain will effortlessly compensate for slightly more delay but will be confused by less).
Future Proof? Worldwide usage?
Sennheiser chose to use the 1900 MHz which is currently relatively free of interference
from competitive devices. This is much less crowded than the 2400 MHz band.
Stand-alone dedicated preamps are usally much better,
especially if you can bypass the aforementioned built-in preamps entirely
(e.g., using a return jack on a mixer).
If you must go through a regular audio input with built-in preamp,
the best rule of thumb to minimize noise is to set it to the lowest available setting
and adjust the external gain so that the meters peak between -18dB and -12dB.
This will keep you safely at no more than 25% of the level beyond which hard clipping occurs.
Panasonic's XLR1 adapter for the Lumix GH5 ($400) :
Although fairly pricey, the XLR1 is a key audio accessory for the GH5 because
it bypasses entirely the regular input amplifiers and replaces them by low-noise ones
which communicate digitally with the camera via the hot-shoe contacts.
The only other way to achieve the same audio quality is to use a good
external recorder to produce an independent soundtrack for later synchronization...
Turn the camera off before connecting or disconnecting the XLR1
to the hot-shoe. (I don't think you could damage the camera by ignoring this
recommendation but this would definitely confuse the software.)
Mounting the unit disables the single 16-bit microphone input of the camera
(unless the user chooses to disable the XLR1 by software).
The XLR1 contains a pair of audio preamplifiers and 24-bit converters.
To save battery power, the XLR1 will turn itself off when the
camera goes into viewing mode. That will turn off the connected
devices which depend on 48 V phantom power or those which merely
sense it, including the Sennheiser AVX
wireless microphone system (after exiting viewing mode, allow 10 seconds
for the AVX system to properly re-establish its radio link).
If you absolutely can't leave with that, give up the AVX auto-off feature
by not feeding it phantom power at all (this will force you to turn
the AVX receiver on and off manually).
The two audio tracks (left and right) of a video are simply not enough
to solve all recording situations.
In some cases, it's not even an option
(no audio is recorded with variable frame-rate (VFR)
An external digital audio
recorder adds considerable flexibility to common recording situations.
Handy Recorders from Zoom
All of Zoom's "H" handy recorders share the same high audio quality going
from compressed MP3 to CD-quality
44.1 kHz sampling rate)
and video-track standards from 16-bit 48kHz to 24-bit 96kHz.
Chronologically, Zoom introduced the H2 first, in
The popular entry-level H1 was released in
It fits the needs of most video bloggers.
The news from the CES show in Las Vegas (January 2018) is that the H1 is being discontinued and
replaced by a new model, the H1n, which will be widely available this month.
The new H1n retails for $120 while unused H1 units are still available on Amazon for $70,
while they last:
The H1's custom amber LCD is replaced by a bluish (96 by 64) dot-matrix
LCD of the same size (1¼").
The front panel of the H1n has controls
formerly located on the side or rear panels of the H1.
There's now a dedicated dial to control input level.
The H1's rear sliders have been replaced by buttons
whose status is updated on the bottom line of the LCD.
The H1n counter is able to display hundred of hours;
32GB in MP3 is up to 555 h 33 min.
(Early H1 firmware was limited to 2GB files; at most 34 h 43 min in MP3.)
New features include a switchable limiter and several low-cut filters.
The H1n now uses two AAA batteries
(the H1 used one AA). Rechargeable batteries recommended.
They both feature the same proprietary 10-pin extension connector as the H5 and H6 which allows either
two additional microphone XLR inputs (without phantom power) or one of their own microphones
(preferably with an ECM-3
In February 2018, Zoom introduced nominally in their F-series,
a model which might be better classified with their H-series of handy recorders
(but they didn't dare call it H0).
The Zoom F1 is a very
compact digital stereo recorder without any built-in microphone.
Instead, it has the same 10-pin proprietary port as the H5 and H6 and can
accept the same capsules.
Either that or it can use a 3.5mm microphone/line input, like the H1 or H1n
(especially for lavalier mics).
The F1 is available in two different bundles:
(2018-01-07) On the original Zoom H1 ultra-portable recorder:
Tips for setting up and using the Zoom H1 handy recorder.
With the Zoom H1, I highly recommend the APH1 accessory pack
($20) which includes
a nice branded protective case, mini-tripod stand, AC adapter, USB cable,
foam windscreen & conical adapter for standard mic clips.
(The APH1n, for the Zoom H1n, has a different protective case.)
($6.25 a pair,
$6.99 for three)
are a must for the microphone input and/or headphone output, if you want to put the unit in your pocket
(secured with Gaffer tape or an elastic band around the whole thing).
($6.99 a pair)
are an even better and slimmer option. The cables from Cable Creation
are the best as their right-angle slim connectors will be flush with the unit, which is highly desirable.
For the Zoom H1 to generate accurate time stamps, you must first
set the time and date:
This is accomplished by holding the red button before you turn the unit on.
That way, you can access the six successive components of the date and time by pressing the
play button and change the blinking number up or down by pressing either
forward or rewind.
Next thing you have to do is choose your recording format.
Select WAV on the bottom of the unit for uncompressed recording
(compressed MP3 is only useful for cramming many hours of stand-alone sound on the micro SDHC card).
Then pressing either forward or rewind
allows you to select one of 6 choices; the combinations of 3 sampling rates
(44.1kHz, 48kHz and 96kHz) and 2 resolutions (16-bit or 24-bit).
If you intend to create video soundtracks, forget 44.1kHz (this is exclusively for audio-only CDs).
96kHz is a definite overkill, both in theory
and in practice
(you end up discarding the extra information anyway on current delivery platforms).
Therefore, consider only 48/16 or 48/24. I use exclusively the latter mode
myself, for the utmost in quality (lower digital noise, greater dynamic range
and perfect match with the best digital video formats).
This is consistent with the quality of Panasonic's XLR1
audio interface and with Sennheiser's AVX wireless system.
Even if the ultimate goal is to deliver CD-quality sound (16-bit) it's good
to record at a higher resolution. In theory, the extra 8 bits correspond
to an additional 48 dB in dynamic range, for a grand total
of 144.5 dB. The range of some microphones is up to that.
In post-production and later compression, at most a 96 dB
portion will be used, but you rarely know which part it will be...
So, keep some headroom and always record at 24-bit & 48 kHz.
Just make absolutely sure that you never clip while remaining well above the noise floor.
No hiss, no clipping and let the chips fall where they may.
With a 32GB card, the Zoom H1 can record 30 hours 46 minutes and 11 seconds of
audio at 48 kHz in 24-bit resolution. (The 2GB card bundled with the unit is only good for
115 minutes and 23 seconds.)
Once this setup is done, once and for all, the use of the unit is extremely simple and intuitive,
with or without the help of its
(which is actually identical to the official
The only delicate part, as usual, is to properly adjust input levels manually.
There's a substantial degradation in sound quality if you trust AGC,
which is best reserved to special situations (like low-quality recording questions from an entire classroom
with a single fixed microphone).
Besides creating high-quality audio clips in WAV format,
I also use the H1 to record voice memos in highly-compressed MP3
(which still sounds much better than the dictating machines of yesteryear).
Since the unit remembers what type of WAV and what type of MP3 is preferred,
I can switch between the two with the flick of one finger.
When using the files, there's no question about the nature of the contents
because the file types are different.
Note that pressing play during recording creates a mark.
You can jump back and forth to such marks during playback by using rewind
and forward. Each audio file can have up to 99 such marks.
For best results, set the input level manually to 37 or more
(the equivalent input noise is higher at 36 or below). Increase the level until
the audio peaks at -12dB or so on the H1 meters. If you need to go well above 60 to
do so, your input signal is probably too weak. If you need to go below 37,
it's probably too strong. In either case, consider changing the volume of
the input device itself, if you have any way to do so (or else, there's no great
harm in going outside of the optimal 37-59 range).
Out of the box, the H1 input level is set at 50.
With firmware 2.0 and above (2013) the Zoom H1 can also be used as a card reader or a USB
audio interface. Zoom's corresponding supplement to the H1 user manual is
Just connect the unit to a computer via USB before turning it on.
The display will alternate between "Card" and "Audio"; press the red button when
your desired choice is displayed (if no selection in made within 10 seconds,
the unit defaults to a card reader).
In practice, once you put a 32 GB card into this unit (that's the largest SDHC capacity)
and format it, you'll never have to remove it.
Just connect the computer with a USB cable to fetch files and erase them  (or you may format
the card each time you start a new job, by pressing trashcan while turning the unit on).
The H1 derives its power from the USB connector if available and you can run it indefinitely this way,
without ever draining the battery (don't expect it to be recharged, though).
In January 2018, when the H1 unit was being discontinued, the latest version of the firmware was 2.10
(displayed as 2/10 upon any ordinary power-on).
If you need that final (?) firmware update, the relevant update file is
mirrored here. Put that file
(H1MAIN.bin, 852,224 bytes) at the root of a micro SD card and power up
the H1 with that card in it; you'll be asked to confirm that you really want to perform the update.
(See user's manual.)
(2018-02-08) The Zoom H5 four-channel handy recorder:
the Zoom H5
delivers more recording capabilities than most small indy producers need
(just add a couple of cheap single-channel units for those rare occasions where
sound is needed from widely separated sources, or when the H5 itself is deemed too bulky).
The audio quality is great, although some quieter preamplifiers are available.
Phantom power (12V, 24V or 48V) can be brought to either of the two built-in inputs
(female combos accepting either XLR or quarter-inch jacks). The inputs on the optional
which may replaces the standard XY stereo microphone cannot
provide phantom power (they're thus suitable only for dynamic microphones,
self-powered microphones or line-level connections).
(2018-02-12) USB interfaces, USB microphones, sound cards.
(In order of preference.)
These are useful in studio work when direct input into a computer is desired,
without going through a separate sound recorder or a video camera.
A sound card fits inside a computer and provides it with analog audio
inputs. Just like the built-in microphone socket of a laptop,
this has the disavantage of exposing sensitive microphone signals
to the noisy electrical environment of a computer.
By contrast, a USB interface will accomplish exactly the same thing but
perform the delicate analog-to-digital conversion away from the computer.
(Only digital signals enter the computer enclosure.)
A cheap alternative is to buy a microphone combined with a USB interface.
Those so-called USB microphones have entry-level audio quality
and provide neither an upgrade path nor any flexibility of use (it's
even chancy to run just two USB microphones at once).
Also, a USB microphone is useless when you're away from you computer...
The better solution is clearly a USB interface
(yielding as many digital channels as there are analog inputa)
or a small USB mixer
(which combines irrevocably several analog inputs into a single digital channel).
Sound slates used to be called clappers or clapperboards.
Originally, they consisted of hinged sticks nailed on black chalkboards.
Now they're white or translucent
plates on which dry-erase markers are used.
Magnets help the sticks snap shut.
A full-size slate is about 9½" by 12"
(bargain size: 8" by 10").
A smaller version, used for tight shots,
is called an insert slate and measures about
6" by 6", including marker sticks.
There's still no better way to match video and audio
clips recorded separately. Even when
timecodes are used,
an old-school slate helps identify everything.
Below is a summary of how professionals go about it.
Slating is normally the responsibility of the 2nd AC
(second assistant-cameraman) also called clapper-loader,
for that reason. (The 1st AC is the
The second AC also maintain a camera record.
Sound recording is always started first (film used to be expensive).
The word speed is yelled to confirm that sound is being recorded
(in the old days, it used to take a short while for the reels of the recorder to reach
their operational speeds).
If there are several shots in a scene (different angles, different lenses, etc.)
they are identified
by one of 23 letters (skipping I, O and S, which could be mistaken for numerals).
Use two such letters if there are more than 23 shots, starting with AA, AB, etc.
It's best to use the
radiotelephony spelling alphabet
(Alfa, Bravo, Charlie...) or any
thereof (Able, Baker...).
Funny alternatives are sometimes improper.
The information on the slate should be accurate and updated for each "take".
The clapper is always brought into the frame in the open position and the essentials
are read (shot ID and take number, at least) the actual clap is preceded
by the word marker or mark. (Possibly preceded
by the identifier(s) of the camera(s) involved.
E.g., "A mark", "A and B mark", "A and B common marker", etc.)
Some graceful recoveries from slate snafus include:
AFS: After false start. This acronym is placed under the take number
(and pronounced aloud) when a shot is interrupted for any reason and restarted from scratch.
Second stick : The term identifies a second marker performed when a
the camera or an audio recorder missed the first one.
Tail marker : Those words precede a marker done with an upside-down slate
at the end of a shot if there was no marker at the beginning.
Put the slate right-side up after the clap.
When the slate is used just to identify the image (without associated sound)
it's held between the sticks,
In North America (not the UK) such a thing is called
an acronym for one of many equivalent meanings:
The ultimate expression of slate etiquette is the soft stick call,
which is used to indicate that the clapper won't be used with full force out of respect
(for example, the slate may be very close to an actor's face in a quiet scene).
A dog-clicker produces a loud click which trainers use as an audio feedback for dogs and other pets.
It's also a great instrument to speed-up voice-over editing with a DAW.
The trick is to click after each fumble to clearly mark where the fumble ends and the do-over begins.
The sharp waveform of the clicker is easily recognized visually and it has to be removed
from the final cut along with a short segment that precedes it.
The beginning of that segment matches what comes just after the click mark.