Flute Tone Investigations
Appendix 1 - Technical Resources
What equipment and software
are being used in these investigations? What are their
limitations, and are these impacting on the quality of the measurements?
Do they impact on the findings? How good enough is near enough?
We'll start by looking at the hardware that will be
involved in the investigations.
The Recording Microphone(s)
The item most likely to affect the accuracy of our
analyses is the recording microphone, but here, luck is definitely on
our side. As our initial skirmish with measured flute sound, the analysis
of the recording of Paul Davis, showed, we are not expecting to see a
particularly wide range of signals, at least in frequency terms.
The lowest note we could encounter on a C flute is 261.6Hz (C4).
By about 4kHz (around C8), the harmonics are some 40dB down and will be
approaching inaudibility. So if we can smoothly cover the range
250Hz to 5kHz, there won't be much that escapes our interest. The
whole audio band covers 10 octaves, we're concerned with less than half
The mic I'm hoping to use is the
Behringer ECM 8000, sold as a "reference mic" for sound system
installers. It's omnidirectional, based on an electret microphone
capsule, balanced and runs on P48 Phantom Power. Being
omnidirectional, it is a simple pressure sensor, and is not prone to
proximity effect as are the more typical cardioid microphones
encountered in recording and stage work. It should measure equally
well in the diffuse field and inline with the incoming sound.
I'd like to get away with that mic if we can as it is
cheaply and readily available everywhere, making it possible for anyone
else to replicate my results and carry on with their own studies.
Here's a calibration chart for a typical example of this type. For
convenience, I've marked in red on the chart the range of frequencies
we'll be expecting. As you can see, the microphone is
substantially "flat" over that range, just starting to rise a little at
the top end. The degree of rise on this mic was 0.89dB at 5kHz
compared to the midrange 1kHz, so less than the 1dB change that a
musically trained ear is capable of perceiving under ideal listening
conditions, listening to a constant tone. Looks good so far!
Where we might run into problems with the Behringer mic
is when conducting noise or clarity measurements. The small diameter
capsule that ensures a good frequency response particularly in the
diffuse field, and the use of a pre-charged dielectric (an electret mic)
reduces the mic's signal to noise ratio. It's possible that, if we
run into trouble here, we can switch to using other mics, eg the
writer's own microphones based on AKG full condenser capsules.
We'll also be looking into the use of miniature electret
mics to investigate what's actually happening inside the flute, eg at
the stopper, and at the foot. We may find that such mics are not
that accurate, but if necessary they can be calibrated off our reference
I'm not committed to the use of the Behringer mic as our
primary reference; if we find we need to get something better, we will.
The price of a
Type 1 measuring microphone is not out of the question (about one
6-key flute, to express it in an internationally understandable currency).
But such expenditure is likely to put off others.
We will also want to accept and analyse recordings made
by others, and of course here we have no knowledge of or control over
the mic or the systems connected to it. And they are unlikely to
have access to calibration equipment. Fortunately, there is a
simple test we can rely on. The range of frequencies we need to
cover is approximately the same range of frequencies represented by the
human voice. If a recording of a voice plays back realistically,
we can take that as a very good sign that a recording of a flute will
contain no huge surprises. This makes it look very probable that
flute players around the world will be able to participate meaningfully
in these investigations.
I'm not particularly happy about my current microphone
preamplifier - an "Ultragain Pro" "High Precision Tube Mic/Line Preamp
Model MIC2200", also made by Behringer. When it works it seems OK
(but with no remarkable capabilities). Like many companies these
days, Behringer seem unaware that switches for audio need to be based on
sliding contacts using precious metals. (Oxides formed on regular
metals soon render snap type switches unreliable for audio.)
Consequently, the switches on the device cannot be trusted to remain in
contact, and often need to be "exercised" (pushed on and off repeatedly)
to make them work again. Hardly "precision", hardly confidence
inspiring, and rather annoying! I admit I bought it in a moment of
weakness when I was too preoccupied with other matters to do the decent
thing and build my own. I will use it for the time being but the
future control unit will incorporate a pair of precision microphone
preamps to replace it.
D-A and A-D Converters
Once we have decent signals, we need to convert them
from analog to digital so that the computer can deal with them.
The soundcards fitted to computers are "not bad", but far from top
class, and it's the microphone input that is usually worst of all.
So, to do this properly, I wanted to put the Digital to Analog
converters beyond reproach. I've opted for an external unit, a
PreSonus Firebox, connected to the computer by Firewire. It
offers an audio performance far greater than we could possibly need,
especially considering the relatively undemanding nature of flute sound.
I'll obviously be interested in listening closely to
recordings I make, and others I may receive, so speakers better than
those that normally come with a computer would be an asset. I've
Fostex PM0.5 MkII Nearfield Monitors. These incorporate
bi-amped 2 x 35w amplifiers in each unit, and claim a free-field
response of 50Hz to 20kHz +/- 2dB. Of course our measurements do
not rely on the qualities of the loudspeakers, so providing the response
of the speakers is enough to alert us to issues, that's going to be
plenty good enough. Fostex offer a subwoofer to extend the
response below 50Hz, but since we're only interested in 250Hz and above,
that's not needed in this application. Might add one later for
personal gratification in other applications!
After a long career in professional broadcasting,
recording and sound restoration, I'm used to having manual control over
signal routing, levels, etc. I find the "Mickey Mouse" mixers in
computer sound ultimately frustrating. So, I'm planning
construction of a dedicated custom control panel to return full manual
control to me. I've mocked up a crude passive interim device to
test out the concept while I confirm the final design. Even so,
it's a godsend already!
Essentially, the control unit allows me to select the
inputs to the A to D converter, direct the outputs from the D to A
converter, and let me choose to monitor signals coming in or out at
will. To be environmentally responsible, I can also shut down all
the audio (including the powered speakers) when not in use, at a single convenient power switch.
The final control unit will do much more. It will
also incorporate the precision microphone preamps and P48V phantom
power, provide drive for outputs, speakers and headphones, and headset
microphone for Skype, incorporate high impedance inputs for oscilloscope
probes, and provide calibration outputs and measurement to enable
absolute measurement. An interim specification for the control
unit is at Appendix 3. I'll publish more details as the design
Sound Level Meter
We won't have a lot of need for absolute sound level
readings, but I do have an old analog Realistic type II sound level
meter that will probably fill the need. I did have a more modern
digital unit with all the bells and whistles, but it died spontaneously
just after the warranty period, and the suppliers advised that such
units cannot be fixed, at least not economically. That makes it
unattractive to replace it with a similar device from this modern era.
We seem to have entered a dark period in human development!
This was borne out recently by a power meter I bought.
When I plugged in the batteries, it didn't come on, and one of the
batteries quickly became very hot. I removed the batteries and ran
an ohmmeter across the battery terminals - the battery holder was dead
shorted. This device could never have been operated. It
still had the Quality Check (QC) sticker on the front panel to assure me
all was well. Clearly, the QC sticker simply attests that the case
Recording and listening environment
The recording and listening environment is an important
contributor to overall accuracy. Ideally this work would be done
in an anechoic chamber, but we're not looking for the top level of
scientific repeatability, so hopefully we'll get away with an office
well packed with books! We will be conducting tests to determine
to what extent standing waves within the room are influencing
Background noise is another issue that will impact on
recording and measurement. Fortunately, we are located on the
suburban-bush fringe, and urban noise is negligible. The dominant
noise source (when the kids are out and the cat's been fed) is the noise
from the office computer we'll be using to record on to. A new
computer has been ordered, and an ultra quiet power supply specified.
A lined cabinet will be made up to minimise interference from the
Now why should we need test equipment? We just
plug it all together and it goes, right?
As we've seen above, what products offer doesn't always
translates to what they deliver. Sometimes, they simply give up
the ghost, like my Digital Sound Level Meter. Sometimes they
linger, like my Behringer High-Precision Tube Mic/Line Preamp. But
sometimes they seem to work, but give performance far short of what they
promise, and that is perhaps the worst sin of all. As I mentioned
above, we are living though a dark period, where we seem no longer
capable of fixing things we've built. It's become clear that the
prices things are sold at these days are too little to allow the device
to be tested in the factory. The moral of the story is, if we
don't test it thoroughly, we have no idea how well it is performing.
Fortunately, these days, most audio tests can be run
using software designed for the purpose (and detailed below), providing
you have appropriate interface to connect the device under test to the
computer. I've built such an interface device, and a version of it
will be incorporated into the control unit being developed. In
addition, we have multimeters to handle the usual range of DC and AC
current and voltage, and resistance, inductance and capacitance tests.
Artificial flute blower?
I hadn't intended this when I set out on this program,
but I'm finding more and more reasons to consider building an artificial
flute blower. It clearly can't replace humans, but can take them
and their weaknesses temporarily out of the equation while we study the
underlying physics. Such a system would include a low noise,
adjustable, stable air source, a way to hold flutes, and an artificial
mouth with artificial lips, adjustable for gap width and height, and for
angle and distance.
This section deals with the software we'll be using to
carry out the investigations, and to validate and calibrate the systems
to be used. Where possible, I'll try to confine myself to using
freeware and shareware, so that others can try out what I'm doing
without need for expensive software.
Some of this software I'm only coming to grips with
myself, so you might find I shift horses midstream.
is a free Digital Audio Editor created by an international group of
enthusiasts. We should be able to use it for all our recording and
replay purposes, and for editing recordings to remove false notes, etc.
We can also use it to view waveforms. Audacity also incorporates
filters and some analytical capacity, including FFT. It can export
files as .wav and, with the associated Lame plug-in, in the convenient
Scope or Soundcard Oscilloscope is a very nicely presented software
test suite incorporating a two-channel oscilloscope, Fast Fourier
Transform, a two channel signal generator and more. It is not
freeware, but available freely for non-commercial educational purposes.
It should help with fault finding, proof of performance and some
is like Scope but seems to offer even more. The downside is that
it isn't quite so clear in its operation. I suspect it will repay
Rightmark audio analyser is a truly extraordinary test facility for
stereo soundcards. You plug the output of your soundcard back into
the input ("loopback"). After setting the levels, Rightmark
provides a burst of special signals, and analyses what it hears coming
back in, showing the shortcomings of the soundcard. If asked, it
even writes a report in .html, with graphs, tables and evaluations.
Of course it's not limited to testing soundcards.
Incorporate a mic preamplifier in the loopback and it will test that as
well. It even provides a facility for subtracting the shortcomings
of the soundcard test from the combined soundcard and preamp test.
Neat-o! We'll use it to prove our control unit and sound interface
are up to scratch.
Mode Calculator. The dimensions of a room predispose the room
to favouring some frequencies over others, so it's worth checking out.
Calculating room modes is a simple enough job, but this calculator makes
it a snap.
Frequency-distance calculator conveniently lists those frequencies
that might have quarterwave resonances at the distance entered.
Useful for identifying the reason behind any peaks notes in the room
EQ Wizard is a nicely presented Java application for measuring room
responses and correcting modal resonances. It includes tools for
generating test signals; measuring Sound Pressure Level; measuring
frequency and impulse responses; generating spectral decay plots,
waterfalls and energy-time curves; generating real time analyser (RTA)
plots; calculating reverberation times; displaying equaliser responses
and automatically adjusting the settings of parametric equalisers to
counter the effects of room modes. We can use it for tuning our
recording and replay environment. Available for free after joining
the HomeTheatreShack forum.
Realtime analyser is
another very comprehensive suite of audio test software, and another one
that I haven't delved into deeply enough yet. It is not freeware,
but can be tried out for free. It then degrades into limited
functionality until purchased. A feature that might interest us is
the ability to synthesise complex waveforms by adding harmonics in
adjustable amounts - the opposite of FFT. This has the capacity to
help answer questions about how we perceive these harmonics.
There are several other suites similar to the above
which I haven't yet had the time to delve into. If they look
promising, I'll note them here.
If you have come across interesting audio test and
measurement software that appears to have something to offer our
investigations, let us know!
And we shouldn't forget our old standbys that have
served us so well in tuning research:
We may well be turning to them to help explain some of
the correlations between tonal and tuning issues.
We'll need facilities other than sound related.
These are available and should help:
Stereo zoom microscope:
Objective zoom range: 0.7x - 4.5x (zoom ratio 6.5:1)
Wide field eyepiece, WF10x/20mm
Working distance: 100mm - without converter lens
Total magnification: 7x - 45x
Round working stage, 95mmŲ
Coarse focus adjustment knob
Inter-pupillary distance: 54-75mm
Dioptre adjustment: ±5mm
45° inclined, 360° rotatable binocular head, wide fan-shape base, no
LED ring-light with 4 switcheable sectors and
Microscope camera, 3.2 MPixels, High speed, USB2,
with ScopePhoto data acquisition and annotation software.
ScopePhoto also facilitates measurements of distances and angles.
Digital Camera with Macro, Nikon 3.34 MPixels
Digital Scales, 500gm, 0.01gm resolution, with
Temperature and Humidity meter, Testo 608, 0-50C,
+/-0.5C, 10-95%RH, +/-2%RH.
Datalogger, for long term temperature and
Magnahelic Flute Leakage Detector
Manometer for blowing pressure measurement
Measuring tools - micrometer, verniers, rules,
bore gauges, magnifiers, etc
You might wonder if I have the required technical
background to carry out these investigations. After all, what
would a common flute-maker know of precision measurement, high end
professional audio, electronic design and construction, musical
acoustics, software development, etc etc? A fair question.
Like probably most flutemakers, I didn't start out that
way. My early career was in electronics, in the laboratories of
the Research School of Physical
Sciences, and later the Research
School of Earth Sciences, at Australia's most advanced research
university, the Australian National University in Canberra. Our
group designed and built the measurement and control equipment used by
scientists investigating everything from diffusion at temperatures close
to absolute zero, through nuclear research, to telescopes, to the age of
the earth itself. Because the research was cutting edge, the
equipment we designed and built had to be cutting edge. We were
even favoured with moon-rock samples for analysis from the lunar
landings, so we were definitely up there with the world's best.
After that, I decided I wanted to get into something
with more human interaction and musical interest, so I moved into audio
engineering in Australia's fast growing community radio sector.
Over the following ten years or so, I designed and built four broadcast
suites (including a lot of the equipment), two recording studios and two
transmitter sites for two radio stations.
Then I got lured into sound preservation at two of
Australia's top-end sound archiving institutions - starting at the
National Library of Australia, and ending up as Head of Sound
Preservation at the National Film & Sound Archive.
It was only after all that that I felt flute making was
a viable full-time enterprise. I'd been doing it since the mid
1970's but limited by the market for Irish flutes in this relatively
sparsely populated land. But when the Internet really got going,
the world was suddenly my market place.
But science wasn't forgotten. I was lucky to have
one of the world's most respected musical acoustics experts, Professor
Fletcher in the next valley. And I fell in with Professor
Wolfe at the University of New South Wales to help work on computer
assisted flute design. (Both flute players as it turns out.
As someone once said: "What is it with Australians and flutes?")
As you can imagine, I haven't been slow in learning what I can from
So, I think I can say with confidence, yes, I have the
background to pull this off. And the resources, and the
determination. I'm also just a few years off the official
retirement age in Australia, 65, and while retirement is the last
thought, I will be able to divert more of my time to research.
Clearly, there's still much to be done!
It's all rather good isn't it. Our laboratory
equipment and software available either for free or not much - were
researchers ever so favoured?