Iain Gray

Snake Charming—The Musical

Python
Iain Gray
AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

ISBN 978-3-319-60659-0 e-ISBN 978-3-319-60660-6

https://doi.org/10.1007/978-3-319-60660-6

Library of Congress Control Number: 2017947760

© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the

Publisher, whether the whole or part of the material is concerned,
specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other
physical way, and transmission or information storage and retrieval,
electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.

The use of general descriptive names, registered names,

trademarks, service marks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are
exempt from the relevant protective laws and regulations and
therefore free for general use.

The publisher, the authors and the editors are safe to assume that
the advice and information in this book are believed to be true and
accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with
respect to the material contained herein or for any errors or
omissions that may have been made. The publisher remains neutral
with regard to jurisdictional claims in published maps and institutional
affiliations.
Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330
Cham, Switzerland
To Jim Martin, a fellow traveller on the road to Parnassus, this book
is dedicated in friendship
Preface
Intended Audience
This book is an innovative introduction to Python and its audio-visual
capabilities for beginning programmers, a resource for expert
programmers and of interest to anyone involved in music. It is
structured around four extensible, audio-visual projects on music and
sound. The beginner will appreciate the ‘need to know’ basis of the
presentation of Python for each project. Expert programmers will be
able to go straight to the project code, run it and then extend it as
they see fit. Musically interested readers will enjoy the historical and
theoretical material at the beginning of each project, and it may even
tempt them to try some coding—it is not too difficult! The projects are
all self-contained but can be extended to incorporate aspects of the
others. Above all, the book is suited for self-study, which should be
playful (pun intended)!

Prerequisites
A minimal understanding of your computer’s operating system is
assumed. In particular, you should be able to

1. find, create and delete files or folders within the filing system

2. open, close, save and print files

3. run program files by opening or ‘double clicking’ them

4. that’s all!

Typography
Normal text will appear in a Times Roman font, whereas Python
code will appear in a monospaced courier font. The first use of a
new program construct will be highlighted as construct . Control
and modifier keys will appear between angle brackets and in a bold
typeface as in < return > or < enter >.

A Note on the Code

The code largely complies with functional and procedural paradigms,
and a reference to object-oriented Python is contained in Appendix.
In the New Language Features sections, it is intentionally repetitive
and uses many print statements, rather than breakpoints, to explain
its operation. The overall driver has been simplicity of exposition.
After all as Albert Einstein said ‘Everything should be made as
simple as possible, but not simpler’.

Structure of Book
In order to achieve expertise in the audio-visual and animation
capabilities of Python, four major projects in music and sound will be
developed. The book is divided into six parts

1. Part I—Installation, shell, editor, Python syntax and package

management

2. Part II—Sound visualisation

3. Part III—Sound creation

4. Part IV—Harmonic visualisation

5. Part V—Composition

6. Part VI—Future development

Part II–Part V comprise the projects and are each split into
background and coding chapters.
 Chapter 14 contains suggestions and guidelines for extending the
projects. Finally, all the internet links and bibliography are in
Appendix.
This book and its projects loosely parallel the historical
development of music; from Rhythm II through Melody III to
Harmony IV, and finally on to Composition V.
Iain Gray
Edinburgh, UK
2017
Acknowledgements
A book like this does not spring fully armed like Athene from the
thigh of Zeus, but relies on the active involvement of several groups
of people

1. my friends and colleagues in Radar Systems Design for

discussions over several decades.

2. my parents for constant support and encouragement.

3. my editors at Springer, Beverley and Nancy, for their courtesy

and patience.
Contents
Part I Snake in the Grass—Python and Its Environment

1 Installing Python

2 The Python Shell—IDLE

2.1 Basic Python Syntax

2.1.1 Comments

2.1.2 Indentation and Block Structure

2.1.3 Input and Output

2.1.4 Declaration of Simple Types and Type Casting

2.1.5 Arithmetic Operators and Precedence

2.1.6 Conditional Expressions, Relational and Logical

Operators

2.1.7 Conditional Statements

2.1.8 Looping Statements

2.2 Entering Python Code

2.2.1 The Python Interpreter

2.2.2 The Python Editor

3 Package Management

3.1 Anaconda

3.1.1 Installing Anaconda

 3.1.2 Using Anaconda

3.2 Alternatives

4 Audacity ®

4.1 Installing

4.2 Using

Part II Banging the Drum—Visualising Sound

5 Mark Kac (1914 to 1984)

5.1 Hearing the Shape of a Drum

5.2 Riding the Waves—Bessel Functions

5.3 Vibrating Plates—Chladni and Germain

5.4 Drumhead Modes

6 Project Code

6.1 New Language Features

6.1.1 Project Header

6.1.2 Plotting Bessel Functions

6.1.3 Graphing in 3D

6.1.4 Animation of Square Wave from Summing Sinusoids

6.1.5 Animating in 3D, for Wave Propagation Along an

Axis

6.2 The Code

 6.2.1 Project Header

Part III Heat in the Desert—Sculpting Sound

7 Joseph Fourier (1768 to 1830)

7.1 The Army of Egypt

7.2 Feeling the Heat—Fourier Transforms

7.3 Chasing Rainbows—Frequency Spectra

8 Bob Moog (1934 to 2005)

8.1 Analogue Additive Synthesis

8.2 Analogue Subtractive Synthesis

8.2.1 Oscillators

8.2.2 Filters

8.2.3 Amplifier

8.2.4 Envelope Generation

8.2.5 Modulation

9 Project Code

9.1 New Language Features

9.1.1 Using Tkinter

9.1.2 Project Header

9.1.3 Sound Storage

9.1.4 Harmonic Analysis

 9.1.5 Oscillators and Mixer

9.1.6 Low Pass Filtering

9.1.7 Implement Butterworth Low Pass Filter

9.1.8 Amplitude Envelope Generation

9.1.9 Low Frequency Oscillator and Modulation

9.1.10 Analysis Displays

9.1.11 Graphical User Interface

9.1.12 GUI Support for Mixer Button

9.2 The Code

9.2.1 Project Header

9.2.2 Oscillator and Mixer

9.2.3 Filter

9.2.4 Amplitude Envelope Generator

9.2.5 Modulator

9.2.6 Amplifier

9.2.7 Displays and Output

9.2.8 User Interface

Part IV The Harmonograph—Victorian Pendulum Toy

10 Hugh Blackburn (1823 to 1909)

10.1 Motion of a Damped Pendulum

 10.2 Blackburn’s Double Pendulum

10.3 Harmonic Ratios—The Lateral Harmonograph

10.4 Parallels—Bowditch and Lissajous

10.5 Of Gears and Motors—The Pintograph

11 Project Code

11.1 New Language Features

11.1.1 Lissajous’ Figures

11.1.2 Damped Orthogonal Pendulums

11.1.3 Harmonic Ratios as Fractions

11.1.4 User Interface

11.1.5 Project Header

11.2 The Code

11.2.1 Project Header

11.2.2 Orthogonal Polynomials

11.2.3 User Interface

Part V Counterpoint à la Mode—Composing Music

12 Johann Joseph Fux (1660 to 1741)

12.1 Gradus Ad Parnassum—Counterpoint

12.1.1 Melody—Direct, Contrary and Oblique Motion

12.1.2 Harmony—Consonance and Dissonance

 12.1.3 Species Counterpoint

12.1.4 Modal Music

12.2 Strict Rules Allow Freedom of Composition

13 Project Code

13.1 The Colours of Noise

13.2 New Language Features

13.2.1 Project Header

13.2.2 Sound Storage

13.2.3 MIDI—Musical Instrument Digital Interface

13.2.4 Composing the Cantus Firmus with Pink Noise

Generation

13.2.5 Sound Generation

13.2.6 Sound Output of Cantus Firmus

13.2.7 The Rules in Python

13.2.8 Output of Whole, Half and Quarter Notes for

Counterpoint

13.2.9 User Interface and Sound

13.3 The Code

13.3.1 Project Header

13.3.2 Generating the Cantus Firmus with Pink Noise

13.3.3 Cantus Firmus Dynamics

 13.3.4 Rules of Counterpoint

13.3.5 Converting MIDI to Frequency

13.3.6 Handling Whole, Half and Quarter Notes

13.3.7 Stereo Output

13.3.8 User Interface

Part VI On Safari

14 Where Next?

14.1 Generic Header

14.2 Part II—Visualising Sound

14.3 Part III—Creating Sound

14.4 Part IV—Visualising Harmony

14.5 Part V—Composition

Curriculum Vitae

Appendix: Internet Links

Part I
Snake in the Grass—Python and
Its Environment
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_1

1. Installing Python
Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

“Latet anguis in herba” Virgil, Eclogues III.

Python is a simple and elegant language that is easy to learn and
install on any platform. The language is cross-platform and truly
multi-paradigm embodying functional, imperative and object-oriented
features. A simple development environment IDLE is also installed
with Python allowing interactive programming and syntax aware
code editing. Sophisticated package managers allow easy access to
pre-made scientific and graphics libraries amongst many others.
On visiting the Python homepage https://www.python.org you will
see.
Clicking on the ‘Downloads’ tab followed by ‘All releases’ in the left
hand sidebar you will then see
so click on ‘Download Python 3.5.2’ to have a graphical installer
loaded on your machine. Open this installer by double clicking on it
and that is all—light blue touch paper and retire! The homepage
identifies your target architecture and will install the relevant version
of Python, and as the language is machine independent Python
source code will run on any supported architecture. Note that Python
3.5.2 was the latest version at time of press but the installation
procedure above will always obtain the latest version.
Finally clicking on the ‘Documentation’ tab followed by ‘Docs’ in
the left hand sidebar will bring you to the documents page
from where you can access tutorials and the main language
references online.
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_2

2. The Python Shell—IDLE

Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

IDLE, the Integrated Development and Learning Environment, is a

shell that comes bundled with the Python distribution on download.
Locate the Python 3.5 folder, open it, find IDLE and open it
seeing a large empty window headed by

the final prompt is where you can enter code interactively for the
Python interpreter.
If on any platform you see the following warning message
it is safe to ignore this and type <return> or <enter> carrying on with
the rest of the chapter. You will only be using ‘Tcl/Tk’ within the
‘spyder’ environment of Chap. 3. To clear this warning you have to
install the latest Tcl/Tk 8.5 release which was 8.5.18.0 at the time of
writing. Visit the ActiveState Tcl/Tk download page at http://www.
activestate.com/activetcl/downloads and scroll down to

from where you can download the latest 8.5 version to install for
your platform.

2.1 Basic Python Syntax

Python itself has an ‘official’ tutorial accessible as described in
Chap. 1 or as in Chap. 3. The aim here is slightly less lofty and is
intended as a ‘get you started’ subset of Python. More advanced
concepts such as functions or datatypes such as lists will be
introduced in the projects as and when required.

2.1.1 Comments
Comments to be ignored by the interpreter are introduced by the
(hash) character (<option> or <alt on many keyboards). Multi-
line comments can be delimited (before and after) by ”’ (three single
quotes).

2.1.2 Indentation and Block Structure

Python is block structured but uses code indentation, rather
begin...end, to delimit related blocks of code.

2.1.3 Input and Output

The most basic functions are input() and print().

2.1.4 Declaration of Simple Types and Type Casting

Only the numeric type of int and float will be used here: an int holds
an integer value and has no decimal point: a float hold a floating
point value and has a decimal point. The two types have different
internal representations. Before being used an alphanumeric
variable must be declared to be of a particular type by

1. assigning it to a constant as in a = 0.0

2. making multiple assignments as in a, b, c = 0.0, 1, 2

3. assigning it to the value of an expression as in a = b/c, where b

and c have previously been declared.
Variables of one type may be cast as the other by using the int()
and float() functions.

2.1.5 Arithmetic Operators and Precedence

The basic arithmetic operators are supplemented by ** for
exponentiation, // for floor (integer) division and for remainder.
Parsing of arithmetic expressions is carried out left to right obeying
the following order of precedence

1. (highest) ** and unary minus but see note below

2.

3. (lowest)
Note that ** is higher than left unary minus, ** is lower than a right
unary minus in terms of precedence.
This order of precedence may be changed by using parentheses
().

2.1.6 Conditional Expressions, Relational and Logical

Operators
Python has a cast of ‘all the usual suspects’ for its relational
operators, which are

1. equals. Note that a single is only used for declaration or

assignment

2. not equals

3. > greater than

4. greater than or equals

5. < less than

6. less than or equals.

These all return Boolean values of True or False and are ranked
below ‘ ’ in order of precedence. They can be chained together to
form multiple relational expressions such as .
Logical operators can be used to combine such Boolean values
into composite conditional expressions with a value of True or False.
They rank lower than the relational operator and have the following
order of precedence
1. not expr—True if expr is False, False otherwise

2. expr1 and expr2—True if both expr1 and expr2 are True, False
otherwise

3. expr1 or expr1—True if either expr1 or expr2 is True, False

otherwise.
Conditional expressions can thus be made of one, or more,
relational expressions tied together with logical operators.

2.1.7 Conditional Statements

Multiple line statements are much more easily entered within an
Editor shell (see later section).
Conditional statements are those which allow the selection of a
consequence or alternative statement depend on the evaluation of a
conditional expression. They come in three flavours which are best
illustrated by examples.

1. conditional statements have a simple structure and are most

often used for boolean assignments, thus out=x and not y
or not x and y

2. inline if is used for conditional assignment and has a simple

structure, thus abs=−x if x 0 else x

3. if, elif, else conditional statements will perform blocks of

statements, after each:, dependent on the result of the
conditional expression. elif short for else if allows for nested
conditional statements. As described earlier indentation is used
to distinguish blocks. For example
if number 0:
print(’negative’)
elif number 0:
 print (’positive’)
else:
print(’zero’)

2.1.8 Looping Statements

Multiple line statements are much more easily entered within an
Editor shell (see later section).
Looping statements are those which allow the controlled
repetition of a statement or group of statements. They again come in
two flavours which are best illustrated by examples. The examples
are in the Python editor section and cover both simple versions of
for loops and while loops. No attempt has been made to explain list
types, or in (membership) operators as such details are deferred to
later Project Code chapters.

2.2 Entering Python Code

IDLE supports two types of shell

1. the Python shell which support interactive development via the

Python interpreter

2. one or more Editor shells which allow you to edit and save
Python code.

2.2.1 The Python Interpreter

Type the following into the IDLE Python interpreter shell, (comments
are optional), followed by <return> or <enter> for each line and
observe the output.
multiple declarations
a, b=355,113
multiple assignments
p,q,r=a/b,a//b, a%b
multiple prints
print(p,q, r)
 s=float(a)+r/b
print(s)
p==s
generally unsafe to use == (equality) on
floats
Boolean declarations
x,y,out=False,True, False
’exclusive or’ as a conditional expression
out=x and not y or not x and y
’abs’ as an inline if statement
abs=−x if x 0 else x

2.2.2 The Python Editor

Unfortunately the Python interpreter shell vanishes on quitting IDLE.
to recall your work you must create a new editor shell, type in your
program and save it with a .py extension. Then on restarting IDLE
open your file to run it, or ‘run module’ in the Python shell via the
Run menu in the Editor shell. The editor offers many features like
automatic syntax highlighting, auto-completion of language words,
auto-indentation of blocks and highlighting parenthesised
expressions.
Open a new Editor shell, and type the following into the IDLE
editor shell, (comments are optional). Save your work in ‘test.py’,
reopen as described above and observe the output. Note that
although strings are used this is not a ‘hello world’ program
read in number
number=0
number=int(input(’number=’))
if, elif, else conditional statement
if number 0:
print(’negative’)
elif number 0:
print (’positive’)
else:
print(’zero’)
list declaration
 sol_fa=
[’do’,’re’,’mi’,’fa’,’sol’,’la’,’ti’,’do’]
note=’do’
simple for loop
for note in sol_fa:
print(note)
feel free to sing along
c=10
simple while loop
while c>o:
print(c)
c=c−1
print(’blast off’)
Finally the syntax highlighted IDLE editor window for ‘test.py’
looks like
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_3

3. Package Management
Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

Over the years the Python community has developed many library
packages to expand and enhance the basic Python implementation.
Of most relevance for the projects in this book are the following three
packages

1. NumPy—a specialised numerical package adding random

numbers, Fourier transforms and linear algebra over multi-
dimensional matrices

2. SciPy—a stack of packages supporting scientific applications

3. Matplotlib—support for two and three dimensional graph plotting

and visualisation
Note that packages are also referred to as libraries and modules
elsewhere, but the generic term package will be retained throughout
this book.
 As well as the packages it is very useful to have a package
manager to automatically install them and keep them up to date.

3.1 Anaconda
Anaconda provides the most comprehensive package manager
currently available for Python as well as giving direct access to
supporting applications and documentation. All the packages are
automatically installed and updated by Anaconda.
During the production of this book Python was upgraded from
version 3.5.1 to version 3.5.2, Anaconda automatically upgraded on
being started up.

3.1.1 Installing Anaconda

Installing Anaconda and the Python Packages could not be simpler,
visit https://www.continuum.io/downloads where you will see
 Scroll down to your computer platform and choose the Python 3.5
installer, preferably if a graphical installer is available use it. Follow
the on-screen instructions, light blue touch paper and retire! You will
then have an Anaconda folder containing all the packages and a
Navigator used to access everything in this folder.

3.1.2 Using Anaconda

Anaconda is launched by opening the Navigator to display its home
screen

Note that some applications may have to be installed before

being used: simply press their install buttons. The home screen will
let you access documentation including tutorials via the left hand
side tabs. For the projects launch Spyder and you will see
 a large syntax aware editor window on the left, an object
inspector on the top right and an interactive IPython window in the
bottom right. IPython is an improved IDLE with parenthesis
matching, automatic completions and many other enhancements,
but note that on a Mac instructions which use the <control> key
should use the <command> key instead.

3.2 Alternatives
Although Anaconda provides comprehensive package management
for Python your needs may be more specialised or limited and
Anaconda is like ’using a sledgehammer to crack a nut’! As honest
advertisers and opinion pollsters should say: “there are other
products (or parties) available”. These include, but are not limited to

1. PyQt—a binding of Python to the multi-platform Qt user

interface

2. PySide—another binding of Python to the multi-platform Qt user

interface
3. wxPython—a binding of Python to the wxWidgets C++ user
interface
4. Python(x.y)—a scientific and engineering packaging of Python
The PyQt GUI already comes bundled with Anaconda. Please be
aware that you may probably have have to manually upgrade any
included packages within them: caveat emptor.
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_4

4. Audacity®
Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

Unfortunately the current ‘state of the art’ in Python precludes

simultaneous sound output, but the use of intermediate ‘.wav’ files
facilitates analysis, recording and manipulation of these sound files.
A free, open source program Audacity is very mature and stable (for
all platforms) and thus is recommended for sound playback.

4.1 Installing
Go to http://www.audacityteam.org and download Audacity for your
platform, taking care to observe any platform specific notes and the
release notes. The manual, containing a tutorial, is also accessible
from the home page.

4.2 Using
On opening Audacity you will see
 and as you are only wanting to play ‘.wav’ files press the in the
top left corner. Audacity provides many more sophisticated tools for
multi-track recording, analysis and playback which are outwith the
scope of the current book.
Audacity® is a registered trademark of http://www.audacityteam.
org.
Part II
Banging the Drum—Visualising
Sound
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_5

5. Mark Kac (1914 to 1984)

Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk
 Mark Kac was a Polish mathematician specialising in probability
theory, who did notable work with Pal Erdös and Richard Feynman.
Today he is most often remembered for an acoustical paper on
hearing drum shapes.

5.1 Hearing the Shape of a Drum

In his seminal 1966 paper “Can you hear the shape of a drum?” Kac
posed a question which was only resolved, in the negative, two and
a half decades later in 1992. The paper delightfully explores the
mathematics required to extract the timbre (tone colour or frequency
spectrum) of an instrument with different boundary shapes
(drumheads), purely from hearing it play a note. As the paper talks
more about tambourines than drums, a more accurate, but slightly
less catchy title, would have been “Can you hear the shape of a
membranophone?”!
The question then boils down to do unique drumheads have
unique timbres, or are there any cases where different shapes
produce identical (isospectral) timbres? This question was
immediately answered, in the negative, by the existence of a pair of
differently shaped drums in the sixteenth dimension which were
isospectral, but this was rather hard to visualise! However in 1992
Webb and Wolpert constructed the following pair of isospectral
drums in the more familiar second dimension.
5.2 Riding the Waves—Bessel Functions
This section which is more mathematically challenging may be safely
skipped, at least on first reading, without losing the theme of this
chapter.
In one dimension waveforms can be described by Fourier
analysis as being composed of the sum of sinusoids, at integral
harmonic intervals as discussed in Chap. 7. This is true for
instruments such as trumpets, violins and clarinets.
With percussion, such as drums, wave propagation over a two
dimensional membrane is governed by Bessel’s differential equation,
. Solutions of this equation for a value of
are known as Bessel functions of the first kind and can be used
to model the modes of vibration of a thin circular, or annular,
membrane such as a tambourine or head of a drum. Bessel
functions of the first kind for are plotted below.
5.3 Vibrating Plates—Chladni and Germain
In the late 18th and early 19th century the German physicist Ernst
Chladni, the “father of acoustics”, conducted a series of experiments
to determine the modes of vibration of a resonating metal plate. The
experiment consisted of a centrally mounted thin square metal plate
with fine sand scattered over its upper surface. This could then be
bowed to resonance and fingers placed at points around the edge.
The sand would settle in areas of no movement indicating the
various modes of vibration of the plate.

Even today Chladni figures are still used in the design of sounding
backs of stringed instruments such as a guitar.

On seeing Chladni’s demonstrations in Paris Napoleon offered a

prize for the best mathematical explanation of the vibrating patterns
observed. The prize was eventually won, at the third attempt, by
Sophie Germain, a self taught mathematician: being a woman she
was barred from attending the École Polytechnique or even from
collecting her prize in person. Despite friendships, by
correspondence, with J.L. Lagrange, A.M. Legendre and C.F. Gauss,
the ’prince of mathematicians’, it was only through her friendship with
J. Fourier, the subject of Chap. 7, that she was able to attend
sessions of the Academy. All this prejudicial treatment might provoke
a latter day Cicero to exclaim “O tempores, O mores”: “what times,
what customs”!
Although this section has been an aside from the main theme of
this chapter it has illustrated interesting parallel work being done on
two dimensional elasticity.

5.4 Drumhead Modes

Returning to thin elastic circular membranes which have wave
propagation governed by solutions to Bessel’s differential equation.
The harmonics are no longer sums of integer multiples of the
fundamental frequency and so euphonic (see Chap. 7). Instead
drumhead modes are composed of multiple inharmonic partials, non-
integral multiples of the fundamental frequency, and so are
cacophonic.
In two dimensions these drumhead modes look like as contour
plots

Whereas in three dimensions, with height exaggeration, they look

like
Animation of these modes is by a double buffering technique. Two
identical Buffers

1. the Display buffer is a frozen version of the image currently

being displayed

2. the Drawing buffer is a dynamic version of the next image

currently being drawn
along with a pair of pointers to each buffer. On completion of the
drawing the two pointers are swapped over. This makes the current
Display buffer the previous Drawing buffer, and the new Drawing
buffer the previous Display buffer. Compare this with a single buffer
is being updated while it is being displayed, as in the Victorian
Zoetrope, resulting in a much more distracting, flickery, displayed
image.
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_6

6. Project Code
Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

Matplotlib, bundled with the Python distribution, was originally a two

dimensional plotting package offering graphing facilities comparable
to its commercial rivals. Recent releases have enhanced this by
adding three dimensional graphics and animation capabilities. This
project will develop the graphics and animation facilities required to
produce an animated circular drumhead as in Chap. 14.

6.1 New Language Features

Project Headers given in New Language Features are applicable
only to this section.

6.1.1 Project Header

You will need to use some features of NumPy, SciPy and Matplotlib
for this project so enter here
import numpy as np
from scipy import special
import matplotlib.pyplot as plt
 Where the lines mean

1. Short alias for package NumPy

2. Extract package special from SciPy

3. Short alias for package pyplot within Matplotlib.

6.1.2 Plotting Bessel Functions

As a simple example of graphing functions type into ‘bessel.py’
import numpy as np
import scipy.special as spe
import matplotlib.pyplot as plt
plt.ion()
plt.grid()
x=np.linspace(0,20,1000)
for alpha in range(3):
y=spe.jv(alpha, x)
plt.plot(x, y, ’-’)
plt.show()
Where the highlighted lines show

1. switch on interactive plotting, after entering %matplotlib in

iPython

2. plot basic grid

3. set up linearly interpolated x axis

4. set up y points as Bessel functions (first kind) for various

5. plot (x, y) with solid line

6. show plot.
the resulting figure should be compared with that shown earlier

6.1.3 Graphing in 3D
Type into ‘rainbow.py’
import numpy as np
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
plt.ion()
fig = plt.figure()
axes = Axes3D(fig)
x = np.linspace(-np.pi, np.pi)
y = np.linspace(-np.pi, np.pi)
x, y = np.meshgrid(x, y)
z=np.cos(x)*np.sin(y)
axes.plot_surface(x, y, z, rstride=1, cstride=1,
cmap=plt.cm.rainbow)
plt.show()
 Where the highlighted lines are

1. Use the Axes3D function within the mplot3d toolkit

2. setup axes

3. linearly interpolate x values

4. setup a grid over x and y values

5. plots surface with row and column spacings and using the
rainbow colour map.
The resultant figure is

There is a good example of drawing a circular drumhead in the

scipy.special documentation referenced in this book backmatter.
6.1.4 Animation of Square Wave from Summing
Sinusoids
Although Matplotlib provides basic double buffering animation
through its interactive functions

1. ion() allowing every plotting operation to update a figure

2. ioff() disabling every plotting operation from updating a figure

3. draw() forces a figure to be redrawn.

There is also an animation module whose most useful functions
are

1. TimedAnimation updates the animation every time interval

specified in milliseconds

2. FuncAnimation updates the animation every specified

function call.
This program illustrates additive synthesis as in Chap. 8 as a
simple 2D animation without redrawing. Type into ‘squsin.py’
import numpy as np
import matplotlib.pyplot as plt
fig, ax=plt.subplots()
x=np.linspace(0,2*np.pi)
harmOdd=[1,3,5,7,9,11,13,15,17,19,21,23,25]
def synthUpdate(oddHarm):
sumSin=0
cnt=oddHarm
while cnt>0:]
sumSin=sumSin+np.sin(x*cnt)/cnt
cnt=cnt-2
return sumSin
for harm in harmOdd:
y=synthUpdate(harm)
 ax.plot(x, y)
plt.pause(1)
Where the highlighted lines are

1. list of the odd harmonics

2. synthesise the odd harmonics by summation

3. overdraw the figure

4. pause for one second N.B.the documentation for plt.pause

has the following caveat “This function is experimental; its
behaviour may be changed or extended in a future release.” It
can always be replaced by time.delay in this case.
The resultant figure is
6.1.5 Animating in 3D, for Wave Propagation Along an
Axis
Using basic double buffering, clearing before redrawing a buffer, and
the mplot3d toolkit, a very effective animation of wave propagation
can be created. Type into ‘propgn.py’
import numpy as np
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
fig = plt.figure()
ax = fig.add_subplot(111, projection=’3d’)
x = np.linspace(-np.pi,np.pi, 50)
y = np.linspace(-np.pi,np.pi, 50)
x, y = np.meshgrid(x, y)
ax.set_zlim(-1, 1)
axes=None
for phi in np.linspace(-np.pi,np.pi, 50):
if axes:
ax.collections.remove(axes)
z=np.cos(x+phi)*np.sin(y)
axes=ax.plot_surface(x, y, z, rstride=2,
cstride=2,cmap=plt.cm.rainbow)
plt.pause(0.01)
Where the highlighted lines show

1. Setup a 3D subplot

2. Setup z limits

3. Allow first frame to be drawn

4. Cleat buffer
5. Pause for ten milli-seconds N.B. the documentation for
plt.pause has the following caveat “This function is
experimental; its behaviour may be changed or extended in a
 future release.” It can always be replaced by time.delay in
this case.
Remember %matplotlib in IPython.

6.2 The Code

Although there is no separate project code for this Part the following
generic header should be used for any of the extensions from
Chap. 14.

6.2.1 Project Header

In ‘main2.py enter
from math import *
from fractions import Fraction
import numpy as np
import scipy.special as spe
import matplotlib.pyplot as plt
import matplotlib.animation as anim
Part III
Heat in the Desert—Sculpting
Sound
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_7

7. Joseph Fourier (1768 to 1830)

Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

Joseph Fourier (no contemporary photographs exist) a French

mathematician and physicist, was recruited by Napoleon as scientific
advisor to his ‘intellectual’ Army of Egypt, initially to survey a
possible Suez canal. Exposure to the desert heat led to a lifelong
obsession with heat and the development of Fourier Analysis.
7.1 The Army of Egypt
What on earth was Napoleon doing campaigning in Egypt and Syria
between 1798 and 1801, when his normal battlefields were in
Europe? The answer is simple: to threaten British trade interests with
the Indian subcontinent. Of more lasting importance was his
recruitment of over a hundred intellectuals to pursue scientific and
archaeological studies in the Middle East. Amongst the discoveries
of this ‘intellectual army’ were two that still resound today

1. the unearthing of the Rosetta stone and the subsequent

decipherment of hieroglyphics by Champollion.

2. the study of heat transfer by Fourier leading to the development

of Fourier Series and the Fourier Transform which are central to
signal processing.

7.2 Feeling the Heat—Fourier Transforms

After being abandoned in the desert by Napoleon’s return to France,
Fourier eventually reached Paris in 1801. So convinced had he
become of the curative effects of heat, that he kept his apartments
unreasonably hot and was himself wrapped in blankets all day.
For the next six years he worked on the propagation of heat
through solid bodies, developing Fourier analysis and Fourier
transforms in the process. His basic tenet, that any repeating
waveform could be expressed as a sum of a series of individual
sinusoids, was initially received with some scepticism, even hostility,
among his intellectual peers of that time. Indeed it was many years
before the monumental and ground-breaking nature of his work was
fully appreciated. As an aside to this work Fourier was the first to
recognise and publicise the ‘greenhouse effect’.
Finally because of his lifelong obsession with heat, he tripped
over his blankets, fell down stairs, was put to bed and died a few
days later. So be warned!
7.3 Chasing Rainbows—Frequency Spectra
Nowadays, rather than in heat propagation, the Fourier transform is
used more often in audio, radio and Radar as a convenient
mechanism for converting time domain signals into the frequency
domain and, using the inverse transform, of converting the frequency
domain into the time domain. For instance in audio it allows the
extraction of the harmonic spectrum of a musical instrument playing
a note.
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_8

8. Bob Moog (1934 to 2005)

Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

Copyright 2017, The Bob Moog Foundation

Seen here with two of his most famous instruments, the Moog
Modular and the Minimoog synthesisers, Dr. Robert ‘Bob’ Moog was
an American pioneer in the design of analogue synthesisers whose
many innovations have become the industry norm of today. From the
early 1950s when he designed and built Theremins, a unique
instrument which you play only by moving your hands without
touching it, his developments included voltage controlled oscillators,
filters and amplifiers, envelope generators for controlling the musical
dynamics and low frequency oscillators for modulating the sound. All
of these are expanded in the section on subtractive synthesis.

8.1 Analogue Additive Synthesis

Although from the foregoing analysis in Chap. 7 it might appear that
the obvious way to synthesise an instrument playing a note is to sum
together its constituent sinusoids this met with several difficulties

1. percussive instruments have many inharmonic (non-integral)

partials as in Chap. 5 which do not conform to standard Fourier
analysis. A pealing bell, for example, has eleven partials before
it begins to sound realistic.

2. timbres of certain musical instruments, particularly the brass and

string sections, require a very large number of high harmonics to
sound realistic.
Although both of these problems can be alleviated by the use of
sophisticate digital signal processing in the analogue world a
different approach was required.

8.2 Analogue Subtractive Synthesis

Although the photograph at the beginning of the Chap. 8 seems a bit
like an antiquated telephone exchange there are really only three
main modules, two supporting modules and a keyboard to consider.
A very good contemporary instrument is the Moog Sub 37
Copyright 2017, The Bob Moog Foundation
The alternative approach taken was to use a small number,
usually two or three, harmonically rich waveforms, filter them to
remove unwanted harmonics and finally amplify them to produce the
note. An envelope generator is used to add dynamics to the output.
Low frequency oscillators are used to modulate the three main
modules to give effects such as vibrato and tremolo. Finally a
keyboard supplies the frequency and duration of the note played.

8.2.1 Oscillators
Only three of the waveforms depicted here, the sawtooth, square
and triangle, are used for oscillator output. All five are used for the
low frequency oscillator in the Modulation section. The input to an
oscillator is a frequency in Hertz (Hz) which is the number of
complete repetitions of the waveform in a second. All the waveforms
and their spectra depicted below have a fundamental frequency of
100 Hz, and were performed at a sample rate of 44.1 kHz.
The figures were all produced using Faber Acoustical’s
SignalSuite and SignalScope software.
Sawtooth oscillators have a full set of harmonics and are are
particularly good at forming the basis for the bright sounds of
brasses or bowed strings.
Square wave oscillators have only odd numbered harmonics and are
ideal as the basis for clarinets. For oboes and beyond pulse width
modulation must be performed to transform the square into an
increasingly rectangular waveform.
Triangle wave oscillators again have only odd numbered harmonics
but these decay at a far faster rate than they do in the square wave.
They are most suited for flute and thin reed sounds.
Notes to observant, and hopefully still interested, readers

1. the first waveform was a ramp rather than a sawtooth - mea

culpa! However their spectra are identical as there is only a
phase-shift of 180 degrees between the waveforms.

2. the spectral folding/aliasing at 22 kHz is an artefact of the

Shannon/Nyquist Sampling Theorem, and does not invalidate
the spectral content being all, or odd numbered, harmonics.

8.2.2 Filters
A filter has a dramatic effect on the timbre of a note. It is most often
a low pass filter which attenuates frequencies sharply above its
cutoff frequency, while leaving those below unchanged. This affects
whether the note sounds warm and dark or cold and bright.
The characteristics of the filters illustrated are

1. large attenuation of signals above the cutoff frequency

2. resonance or emphasis of signals about the cutoff frequency

3. filter slope (Q) indicating the rate of attenuation usually

expressed in -dBV/octave
Note that the cutoff frequency is usually taken as the point when
-3dBV attenuation below the main flat response has occurred.
Butterworth filters will be used in the project code, with
Chebyshev 1 used to introduce resonance as in Chap. 14.
Other types of filters are used less commonly including high
pass, band pass and band stop filters which ‘do exactly as they say
on the tin’. All pass filters do not attenuate frequencies but cause a
phase shift between input and output.
Filters are further classified by whether they involve feedback of
the output into the input as

1. Finite Impulse Response (FIR) filters have no feedback

2. Infinite Impulse Response (IIR) filters involve feedback

Most of the filters used here will be of the IIR type.

8.2.3 Amplifier
The amplifier controls the final output level of the sound which can
be affected by the envelope generators and modulators.

8.2.4 Envelope Generation

The overall dynamics, how the sound builds up and dies away, is
controlled by an amplifier envelope generator.

It traditionally has the four stages shown above

1. Attack phase—the time it takes to reach a maximum output level

from the key being pressed
2. Decay phase—the time it takes to fall to a sustain level
3. Sustain level—the volume the note is held at until the key is
released

4. Release phase—the time to reach zero output level after the key
is released
and is known as an amplitude ADSR envelope generator under the
control of a key on/off gate.

8.2.5 Modulation
Modulation uses a low frequency oscillator in the infra-sonic range
(0–30 Hz) to modify waveforms in the three core modules. Any
waveform in the oscillators section can be used. For example using
a sine waveform and targeting the

1. oscillator gives frequency modulation (FM)—vibrato

2. amplifier gives amplitude modulation (AM)—tremolo

© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_9

9. Project Code
Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

This project will develop the core modules of an analogue

subtractive synthesiser, with audio monitoring of each module and
enable you to

1. change frequency and waveforms of the oscillators, and their

mixtures

2. change the cutoff frequency and filter slope of the low pass filter

3. change the output volume of the amplifier

4. change the attack, decay, sustain and release of the amplifier

envelope generator

5. change the rate and amount of the low frequency modulation

Further time domain and frequency domain displays will be
developed, using Digital Signal Processing, to show respectively the
waveform and spectral content of the output from the oscillator, filter
and amplifier.

9.1 New Language Features

Project Headers given in New Language Features are applicable
only to this section.

9.1.1 Using Tkinter

The graphical user interface (GUI) of the synthesiser will be built
from these widgets within the ‘tkinter’ toolbox which is common
across all platform. Insert the following into the program header

Where these four lines establish

1. tkinter as main package

2. ttk (themed Tk) as the widget package available in Tcl/Tk 8.5

onward

3. colorchooser package

4. root as Tk root level

Within the root level GUI there will be frames supporting

1. the oscillators’ frequency and amount of pulse width modulation

will be determined by list menus, and the mixture between them
by mouse position within a system colour chooser graphical
element
2. the filter’s cutoff and slope will be list menu items
3. overall volume will be controlled by a slider

4. ADSR settings on the amplitude envelope generator will be by

sliders

5. modulation rate and amount will be set by list menus, and its
destination by a radio box (off, vibrato or tremolo)

6. displays for the oscillators’, filter’s and amplifier’s outputs will be

controlled by on/off check boxes

9.1.2 Project Header

You will need to use some features of NumPy, SciPy and Matplotlib
for this project so enter here
import numpy as np
import matplotlib.pyplot as plt

9.1.3 Sound Storage

Arrays from NumPy will be used to hold the numerical samples of
the sound. These will be turned into WAV files for input, output and
offline processing of these sounds. Add the following line to the
project header
from scipy import io

9.1.4 Harmonic Analysis

The manipulation and display of the harmonic content of sounds
requires the use of the Fast Fourier Transform (FFT). Add the
following line to the project header
from scipy import fft

9.1.5 Oscillators and Mixer

This section is long and repetitive because it provides the basis for
sound generation both here and in Chap. 13.
Insert the following Python test program saving it as ‘waves.py’
 whose output is a single wave of the sawtooth(red), triangle(blue)
and square(green) oscillators for repeated storage in their respective
’.wav’ files, and then output for one second, then mixtures of two or
three of these waveforms and finally reading back the waveforms
and mixing them in proportions.
The code is in five sections each headed by a highlighted
comment.
The first section gives an example of a few pyplot commands to
draw the oscillators’ waveforms. plt.plot is followed by two arrays
of x and y coordinates for the line. If only one array is given it is
assumed to be the y coordinates. The string following gives the line’s
colour and appearance and is optionally followed by parameters
describing for instance the line width. Because of the double
buffering technique described in Chap. 5 the output will only be
displayed after the plt.show() function call. This displays a single
waveform of each of the three oscillators.

The second section constructs these waveforms repeating them

freqNote times and writing their output to ‘.wav’ files. There is no
reason why the sampling frequency freqSamp needs to be the
same as the playback frequency but doing so avoids the need for
messy interpolation routines. The CD sampling frequency of
44100 Hz is used, giving a Nyquist (Chap. 8) frequency of 22050 Hz
just above the human hearing range. has 79
factors (see http://factornumber.com/?page=44100) enabling a
choice of integer factorisations for frequency with an even number of
repeats to allow construction of square and triangle waveforms.
These frequencies are
 allowFreqs=
[22050,11025,7350,4410,3675,3150,2450,2205,
1575,1470,1225,1050,882,735,63
0,525,490,
450,441,350,315,294,245,225,21
0,180,175,
150,147,126,105,98,90,75,70,63
,60,50,45,42,35,30]
thus covering the human range of frequencies of 30 Hz 20
kHz. linspace performs linear interpolation over a range for a
number of points, where False says not to include the endpoint. The
variable test1 illustrates the operation. append joins two arrays to
form a longer array. The variable test2 illustrates the operation.
The waveform is built by first forming a single copy using numRpts
and appending freqNote copies to make an one second note. For
square and triangle waveforms the first stage must be performed as
two half waveforms appended together. Finally write puts these out
to a ‘.wav’ file, at a rate of freqSamp using the constructed
waveform data. There is obviously a lot of repetition here which will
be culled out in the final code but as the Bellman says

“what I tell you three times is true”

Lewis Carroll, ’The Hunting of the Snark’

Audacity displays of sawtooth, square and triangle waveforms are

illustrated below.
The third section shows how to mix waveforms in pairs and
normalise them before writing them out. Audacity displays of
mixtures of sawtooth and square, square and triangle and sawtooth
and triangle waveforms are illustrated below.
The fourth and fifth sections show an equal and a proportional mix of
all three waveforms and also how to read back a ‘.wav’ file.
Audacity displays of an equal and proportional mix are illustrated
below, and demonstrate the variety of timbres achievable purely by
mixing alone.

9.1.6 Low Pass Filtering

Filters are used to remove unwanted parts of a signal such as high
harmonics leaving the rest of the signal untouched. A filter taking an
average over a series of time steps provides the basis of a low pass
filter. This can be of two forms, where x(n) and y(n) are the input and
output at time n respectively.

1. Finite Impulse Response (FIR) where the output is purely a

function of the input. A simple example is a moving window
average filter such as

2. Infinite Impulse Response (IIR) is a function of the input and

feedback of previous values of the output. A simple example is
an alpha filter
Insert the following Python test program saving it as ‘alpha.py’
 The code is in four sections each headed by a highlighted
comment.
The first section reads in unmixed copies of the original three
oscillators.
The second section allocates array space for the filtered outputs,
sets alpha for the filter and sets the initial output by assuming
y(-1)=0.
The third section is the alpha filter,
itself.
The final section writes out the filtered values printing them for
comparison purposes. Although the alpha filter is a blunt instrument
for performing keyhole surgery, it has ‘rounded off’ the sharp corners
in the sawtooth, square and triangle waveforms shown below. The
filter cutoff is about 6000 Hz as shown in the triangle spectrum. This
combined effect gives a sound which is mellower, warmer and darker
than the original. More refined filters within the package
scipy.signal will be used in the final code.

9.1.7 Implement Butterworth Low Pass Filter

As an example of the finer control possible implement a twelfth order
Butterworth filter with a cutoff of 5000 Hz, saving it as ‘butter.py’
 Where the five highlights are

1. the Nyquist frequency is half the sampling rate, see Chap. 8

2. for a digital filter the normalisation is [0, 1] where 1 is the

Nyquist frequency

3. the Butterworth design returns a tuple of the numerator (b) and

denominator (a) polynomials of the IIR filter, see Chap. 8

4. computes the initial state for the filter

5. executes the fitter for the given design and initial conditions
the last three stages need to be performed for each new filter type
such as scipy.signal.cheby1.
The Audacity spectrum for this filter exhibits a sharp ’shoulder’ at
the cutoff and should be compared with that from the alpha filter
above.
9.1.8 Amplitude Envelope Generation
The dynamics of an instrument’s sound over time are controlled by
the amplitude envelope generator.
Insert the following Python test program saving it as ‘ADSR.py’
 The code is in three sections each headed by a highlighted
comment.
The first section reads in the filtered versions of the unmixed
oscillators, and sets a time standard of one milli-second.
The second section sets the durations of each envelope
component and the output sustained level.
The third section uses linspace to interpolate the four parts of
the envelope which are then joined using append. This envelope is
then multiplied by the waveform to account for both positive and
negative segments. The output files are shown below for filtered
sawtooth, square and triangle waveforms.

9.1.9 Low Frequency Oscillator and Modulation

A Low Frequency Oscillator (LFO) operates at infra-sonic
frequencies (0 Hz 30 Hz) to produce a signal to modulate one of
the three main modules, the Voltage Control Oscillator, Filter and
Amplifier (VCO, VCF and VCA). With a frequency sampling rate of
44100 Hz any of the following frequencies in Hz are allowed
allowMods=[30,25,21,18,15,14,10,9,7,6,5,3,2,1]
Insert the following Python test program saving it as ’sinemod.py’
 The code is in two sections each headed by a highlighted
comment.
The first section proceeds as before for assembling waveforms,
except that linspace is not used and the sine wave is interpolated
manually. Other modulating waveforms in Chap. 8, sawtooth, ramp,
square and triangle, can all be made with linspace as above.
The second section applies Amplitude Modulation (AM) to the
VCA to produce tremolo. Frequency Modulation (FM) on the VCO
will be discussed in Chap. 14.
The illustrations are of the 5 Hz sine wave and tremolo applied to
the filtered and envelope triangle wave.
9.1.10 Analysis Displays
The analysis displays will allow showing a single waveform (in the
time domain), and its harmonic, spectral content (in the frequency
domain).
Insert the following Python test program saving it as ‘display.py’.

The code is straightforward but several important new concepts

are highlighted. These are
1. scipy.fftpack contains functions which perform the Discrete
Fourier Transform from the time domain to the frequency
 domain efficiently. This transformation uses Fast Fourier
Transforms (FFT).

2. plt.ion sets up interactive graphics on typing %matplotlib

in the IPython window.

3. the next three lines set up multiple subplots within a figure.

4. the next line draws a single waveform using an array slice.

5. the final two lines produce the frequency display using a real
FFT scaled to [0,1].
The final display, with interactive controls for panning and
zooming, looks like

9.1.11 Graphical User Interface

The widgets from tkinter that will form the GUI will be

Overall six Frames to contain each major unit in synthesiser

Overall a Button to start the program when all user entries,
including default values, are completed

Oscillator containing a Combobox to allow selection of

frame frequencies from a list and a mixer Button calling up a
colorchooser to mix the three different oscillators

Filter two Radiobuttons will allow selection of the filter’s

frame order and cutoff frequency

Amplifier a Radiobutton to select output volume level

frame

Envelope four Entry boxes allowing the input of ADSR values in

Generator msec and %
frame

Display a Radiobutton allowing a choice of VCO, VCF or VCA

and outputs to be displayed, and an Entry for specifying
Storage sound storage file
frame

LFO a Combobox to choose the modulation frequency and

frame a Radiobutton to select its depth

9.1.12 GUI Support for Mixer Button

The Mixer will be called from a Button within the VCO Frame. Type
into ‘mixer.py’
from tkinter import *
from tkinter import ttk
from tkinter.colorchooser import *
def getColor():
((red, green, blue), color) = askcolor()
print(’sawtooth = ’, red)
print(’square = ’, green)
print(’triangle = ’, blue)
 print(’color = ’, color)
return color
root=Tk()
ttk.Button(root, text=’Mixer’,
command=getColor).pack()
root.mainloop()
where colour is handled as a nested tuple. Note that the program
is contained in a root.mainloop() GUI event loop it must be
stopped by quitting Spyder.
The mixer Button will call up this palette for mixing
sawtooth(red), triangle(blue) and square(green) oscillators in
proportions. The displays are illustrated from the Macintosh.

9.2 The Code

The project Python code is thus a composite of the above
subsections, with suitable modifications to accommodate the GUI.
Individual sections will be encapsulated in functions, with sound file
data being the parameter mainly used between them. It develops the
three major functional units, of the synthesiser, VCO, VCF and VCA,
and two supplementary units AEG and LFO, along with displays and
sound storage functions.

9.2.1 Project Header

Type into ‘main3.py’
from math import *
import numpy as np
import scipy.io.wavfile as io
import scipy.signal as sig
import scipy.fftpack as fft
import matplotlib.pyplot as plt
from tkinter import *
from tkinter import ttk
import tkinter.colorchooser as col
freqSamp = 44100

9.2.2 Oscillator and Mixer

Within the VCO the oscillators , sawtooth, square and triangle,
provide the sound generation functions, which the mixer the mixer
combines into a single piece of sound data for further processing.

Inputs frequency of note, oscillator mixture

Output sound data for mixed oscillators

Type into ‘main3.py’
oscillators and mixer
def VCO(freqNote=441,mix=0x7F7F7F):
def oscillators():
def Saw():
numRpts=freqSamp//freqNote
sawData=np.linspace(1.0,-1.0,n
umRpts, False).astype(float)
rpts=freqNote-1
sawNote=sawData
 while rpts 0:
sawNote=np.append(sawNot
e, sawData)
rpts=rpts-1
return sawNote
def SquTri():
halfRpts=freqSamp//(2*freqNote
)
firstHalf=np.linspace(1.0,1.0,
halfRpts, False).astype(float)
lastHalf=np.linspace(-1.0,-1.0
,halfRpts, False).astype(float)
squData=np.append(firstHalf,
lastHalf)
firstHalf=np.linspace(-1.0,1.0
,halfRpts, False).astype(float))
lastHalf=np.linspace(1.0,-1.0,
halfRpts, False).astype(float)
triData=np.append(firstHalf,
lastHalf)
rpts=freqNote-1
squNote=squData
triNote=triData
while rpts 0:
squNote=np.append(squNot
e, squData)
triNote=np.append(triNot
e, triData)
rpts=rpts-1
return (squNote, triNote)
sawNote=Saw()
(squNote, triNote)=SquTri()
return (sawNote,squNote, triNote)
def mixer(oscs):
(saw,squ, tri)=oscs
blue=mix%256
green=(mix//256)%256
 red=(mix//65536)
norm=red+green+blue
return
(red*saw+green*squ+blue*tri)/norm
return mixer(oscillators())

9.2.3 Filter
The VCF is the principal sound shaper within the synthesiser
attenuating frequencies above the cutoff at a rate dependent on the
order of the filter.

Inputs sound data, cutoff frequency and order of Butterworth filter

Output filtered sound data

Type into ‘main3.py’
filter
def VCF(soundData, cutoff=5000,order=4):
initialising filter
buffer=np.zeros(freqSamp)
Nyquist=freqSamp/2
Wn=cutoff/Nyquist
(b,a)=sig.butter(order, Wn)
zi=sig.lfilter_zi(b, a)
applying filter
(buffer,_)=sig.lfilter(b,a,soundData,
zi=zi*soundData[0])
return buffer

9.2.4 Amplitude Envelope Generator

The AEG controls the overall dynamics of the synthesiser sound
dependent on an envelope.

Inputs sound data, values for Attack, Decay, Sustain and Release
Output dynamically shaped sound data
 Type into ‘main3.py’
amplitude envelope generator
def AEG(soundData, envelope=(10,10,0.5,10)):
onems=freqSamp/1000
ADSR envelope
(att,dec,sus, rel)=envelope
attack=floor(att * onems)
decay=floor(dec * onems)
sustain=sus
release=floor(rel * onems)
susDur=freqSamp-(attack+decay+release)
4 linspace for envelope
A=np.linspace(0,1,attack,
False).astype(float)
D=np.linspace(1,sustain,decay,
False).astype(float)
S=np.linspace(sustain,sustain,susDur,
False).astype(float)
R=np.linspace(sustain, 0,release,
False).astype(float)
AD=np.append(A, D)
ADS=np.append(AD, S)
ADSR=np.append(ADS, R)
return ADSR*soundData

9.2.5 Modulator
The Low Frequency Oscillator (LFO) allows modulation of the sound
by a low frequency waveform.

Inputs sound data, LFO frequency and depth

Output modulated sound data

Type into ‘main3.py’
LFO modulation
def LFO (soundData, LFOrate=5,depth=0.5):
 sampInt=freqSamp//LFOrate
sampSine=np.empty(sampInt).astype(float)
twopi=2*np.pi
sampSine[0]=0.0
time=1
while time sampInt:
sampSine[time]=np.sin(twopi*time/sam
pInt)
time=time+1
can be replaced by two calls to np.linspace
sineWav=sampSine
cnt=LFOrate-1
while cnt 0:
sineWav=np.append(sineWav, sampSine)
cnt=cnt-1
mod=sineWav*depth
tremolo
return soundData*(1+mod)

9.2.6 Amplifier
The VCA controls the overall volume of the sound output.

Inputs sound data, volume level

Output amplified sound data

Type into ‘main3.py’
amplifier
def VCA(soundData, volume=1):
return soundData*volume

9.2.7 Displays and Output

Time (in number of samples) and frequency (in Hz( displays can be
generated from the sound data from the VCO, VCF or VCA and this
sound data then stored in a file.
Inputs sound data, output file
Output None
Type into ‘main3.py’
display and output
def display(soundData, fileName=’out’):
plt.ion()
plt.figure(1)
plt.subplot(211)
plt.plot(wavData[0:freqSamp])
plt.title(’time’)
plt.grid()
plt.subplot(212)
plt.semilogx()
plt.title(’freq’)
plt.grid()
plt.plot(abs(fft.rfft(wavData))/32768)
io.write(fileName+’.wav’,freqSamp,
soundData)
return None

9.2.8 User Interface

The full synthesiser works with just calling the above functions in a
nested fashion as in
display(VCA(LFO(AEG(VCF(VCO()))))
with or without default parameter values. display can be
inserted before VCO, VCF or VCA depending on the output to be
monitored. A full GUI requires the widgets described above plus the
mixer GUI support function.
Using the default values and %matplotlib in the ipython console
results in
where enlarging the foot of each main spectral line shows the effect
of the 5 Hz LFO modulation, and is probably also an artefact of
performing a real FFT without a window function. The different height
of the harmonics (spectral lines) reflect the mixture of sawtooth,
square and triangle waveforms.
What you can then do is mix different combinations and multiples
of VCF, AEG and LFO in any order, between the VCA and VCO
outermost functions. This gives the sound producing capabilities of
the Moog Modular synthesisers of the 1960s. All this from very
simple synthesis functions!
Part IV
The Harmonograph—Victorian
Pendulum Toy
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_10

10. Hugh Blackburn (1823 to 1909)

Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk
Hugh Blackburn, seen here in caricature, was Professor of
Mathematics at Glasgow University for thirty years from 1849 to
1879. Whilst an undergraduate at Cambridge he designed a form of
double pendulum, described in section below, which bears his name.
This device was used to investigate harmonic intervals.

10.1 Motion of a Damped Pendulum

This section which is more mathematically challenging may be safely
skipped, at least on first reading, without losing the theme of this
chapter.
A simple pendulum, without friction, is a simple harmonic
oscillator and so obeys, to a small angle approximation, the second
order differential equation , where is the angular
displacement of the pendulum, l the pendulum’s length and g is
gravitational acceleration. This gives a period for the pendulum of
by Huygen’s law. This correspondence between length
and period is illustrated by a device called the Pendulum Wave, in
which fourteen pendulums of increasing length are displaced by the
same amount then released, resulting in

This also illustrates the fundamental trigonometric nature of the

solution to the differential equation . Finally to
account for damping caused by friction, at the pivot and air
resistance, simply multiply this solution by a decreasing function of
time such as , where d is a damping factor.

10.2 Blackburn’s Double Pendulum

Blackburn’s double pendulum, from Sound by John Tyndall, 1879.

 Blackburn’s double pendulum is Y shaped. The top part, the arms
of the Y, is generally made of a solid piece of wood, and pivoted so
that it can only move in one axis. The bottom part, the stem of the Y,
is attached to the upper part with some flexible material, like rope,
allowing it the freedom to move in both axes. Attached to this lower
part was a container of sand or ink to record the evolving patterns of
the harmonic ratios.
The major problem with this design is that the pendulum must be
physically very large to record the finer pattern details caused by the
decay of the pendulum motion. As an example see the twelve foot
Blackburn pendulum built by Paul Wainwright in his barn, Please see
book backmatter.

10.3 Harmonic Ratios—The Lateral

Harmonograph
Clearly it was quite impractical to carry around a twelve foot
Blackburn pendulum to Victorian drawing rooms, to entertain the
other guests after a dinner party! The solution came in the form of a
lateral harmonograph small enough to pack in a wooden carrying
case. As well as legs to support the wooden case as a table, it had
two metal pendulums with movable bobs. The pendulums were fitted
in two gimbals at right angles to each other, thus ensuring orthogonal
motion. One pendulum held a drawing surface and the other a
recording pen. The bobs could be moved relative to each other to
establish different harmonic ratios.
The illustrations below are respectively of a beat note, and a
perfect fifth interval, and were made using Francis McConville’s
spreadsheet simulation of a lateral harmonograph, Please see book
backmatter. This site also contains illustrations of lateral
harmonographs.
10.4 Parallels—Bowditch and Lissajous
It is almost a truism in science that the same idea is worked on
independently by different people simultaneously, witness Newton
and Leibniz on differential calculus. In the case of harmonic analysis
Nathaniel Bowditch, the father of ocean navigation, in work later
refined by Jules-Antoine Lissajous, worked on this study
independently of Hugh Blackburn.
Nowadays, when the appearance of Lissajous’ curves of
harmonic ratios is familiar from oscilloscope screens it is very
revealing to see his apparatus from nearly a century before such
displays became commonplace. Two tuning forks were mounted at
right angles to each other with each bearing a mirror on one of their
arms. This arrangement then focussed light on to a telescope
through which the curves could be observed.
These curves are arranged by phase angle against harmonic ratio.

10.5 Of Gears and Motors—The Pintograph

Instead of varying pendulum length, by moving bobs, to set up the
ratios for harmonic intervals, pintographs do this purely
mechanically. This can either be through setting up gear ratios or
varying stepper motor speeds. Resulting in their drawings being
related to, but not identical to, the equivalent harmonograph
drawings.
The pintograph illustrated below is the PrimoGraf from LEAFpdx,
Please see book backmatter. It consists of seven co-prime gears
which can be set up in pairs or triples to achieve different ratios. It is
then cranked manually.
Joe Freedman, www.sarabande.com
The beat note illustrated below, with gear ratios 40:39:41, should
be compared with that from the lateral harmonograph.

Joe Freedman, www.sarabande.com

© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_11

11. Project Code

Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

This project will develop an animation of a lateral harmonograph

where you can specify the harmonic ratios to be plotted. The various
stages involved are

1. Plotting Lissajous’ figures

2. Plotting two orthogonal damped pendulums

3. adding user interface control over harmonic ratios, expressed as

rational numbers, to be plotted

11.1 New Language Features

Project Headers given in New Language Features are applicable
only to this section.

11.1.1 Lissajous’ Figures

Lissajous’ curves are specified by the pair of parametric equations
and where

1. t is time

2. A and B are amplitudes taken as unity here

3. a and b are frequencies, a / b giving their harmonic ratio

4. is the phase offset between the two sinusoids

11.1.1.1 Just Temperament Intervals

Just temperament is used here, instead of Equal temperament
(where all semitones are ), as all the intervals can be expressed
as rational numbers with integral numerators and denominators.
These intervals are

1. Unison, ratio 1:1

2. Octave, ratio 2:1

3. Perfect Fifth, ratio 3:2

4. Perfect Fourth, ratio 4:3

5. Major Third, ratio 5:4

6. Minor Third, ratio 6:5

7. Major Tone, ratio 9:8

8. Minor Tone, ratio 10:9
9. Semitone, ratio 16:15
Enter the following test program into Python saving it as ‘liss, py’.

The first highlight makes the program iterate over cycles,

while the second highlighted for loop can of course be substituted
by a linspace function. For a Perfect Fourth interval (4:3) the
output should look like (after %maplotlib in IPython).
11.1.2 Damped Orthogonal Pendulums
Enter the following test program into Python saving it as ‘damp.py’
 The first two highlights multiply by the damping factor, while the
third highlight iterates over cycles. The diagram shows a
harmonograph of a perfect fifth (after %maplotlib in IPython).

11.1.3 Harmonic Ratios as Fractions

Python expresses rational numbers in the module fractions.
Enter the following test program into Python saving it as ‘ratio.py’
 Where Fraction is the constructor function, numerator and
denominator the accessor functions for Rational numbers.

11.1.4 User Interface

The widgets from tkinter that will form the GUI will be

Overall a single Frame to contain remaining widgets

Overall a Button to start the program when all user entries,
including default values, are completed
Intervals a Combobox will allow selection of harmonic interval
Phase a Combobox will allow selection of phase offset between
the two sinusoids
Plot type a Combobox will allow damping on or off to be chosen

11.1.5 Project Header

You will need to use some features of NumPy and Matplotlib for this
project so enter here
11.2 The Code
The project Python code is thus a composite of the above
subsections, with suitable modifications to accommodate the GUI.

11.2.1 Project Header

Type into ‘main4.py’

11.2.2 Orthogonal Polynomials

Two different pairs of orthogonal pendulums are defined

1. Lissajous an idealised system, without friction, where the

pendulums move at a harmonic ratio and phase shift with
respect to each other

2. Harmonograph a realistic system, with frictional damping, where

the pendulums move at a harmonic ratio and phase shift with
respect to each other

Inputs harmonic ratio as a fraction, phase shift in degrees,

friction as Boolean
Output None
Type into ‘main4.py’
11.2.3 User Interface
A simple GUI is developed to fill in the entries for the pendulums
for more sophisticated GUIs see Appendix A.
Type into ‘main4.py’
 Although this is a relatively simple GUI it illustrates the main
features

1. different types of widget

2. grid geometry manager

3. bindings and control callback on events

4. the event loop

This GUI must be extended with a canvas widget to allow
drawing, see Chap. 14.
Part V
Counterpoint à la Mode—
Composing Music
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_12

12. Johann Joseph Fux (1660 to 1741)

Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

Fux was a prolific Austrian composer of the early Baroque period,

however it is mainly for his music theoretical works on harmony and
counterpoint that he is remembered today. This work covered tonal
music from the 17th to late 19th centuries, the era of Common
Practice. However, as will be shown, this can be adapted to cover
modern jazz and rock genres in the Extended Common Practice.

12.1 Gradus Ad Parnassum—Counterpoint

Fux’s masterpiece Gradus ad Parnassum, steps to Parnassus, was
originally published in 1725 and has remained in print ever since.
Written in Latin, the Lingua Franca of its day, it consists of two parts

1. The Theory—covers intervals as a mathematical exercise,

consonances, motions and rules for motion between intervals.

2. The Dialogue—between the master Aloysius (Palestrina) and

his student Josephus (Fux) consisting of a dozen didactic
problems per species. These problems cover prescribed cantus
firmus above and below the accompanying melody line, in each
of the permissible modes, and can be considered the Baroque
equivalent of cryptic crossword puzzles.
The book was known to have been greatly admired by J.S.Bach,
and that composers like Mozart and Beethoven, who worked out all
the exercises, then influenced the whole of tonal music through its
teachings.

12.1.1 Melody—Direct, Contrary and Oblique Motion

Motion can be considered the horizontal (temporal) aspect of the
music, and of how two (or more) such melody lines move with
respect to each other.

1. Direct motion—both melodic lines ascend or descend in pitch

during the same time interval

2. Contrary motion—one melodic line ascends while the other

descends in pitch during the same time interval
3. Oblique motion—one melodic line ascends or descends in pitch
while the other melodic line remains at the same pitch during the
 while the other melodic line remains at the same pitch during the
same time interval
Fux presents four rules for allowable motions, however these can
be simplified into a single rule—the only forbidden motion is direct
into a perfect interval.

12.1.2 Harmony—Consonance and Dissonance

Harmony can be considered the vertical aspect of the music and
what notes are viewed as being consonant or dissonant to each
other. The consonance and dissonance interpretation will vary
dependent on the genre or period under consideration, for instance
Baroque or modern jazz.

1. the unison, major third, perfect fifth, major sixth and octave are
considered consonant intervals

2. all other intervals, including the perfect fourth, are considered

dissonant and only permitted as transitions between consonant
intervals which resolve them
Transition rules and handling of dissonances will be covered in
detail in Chap. 13.

12.1.3 Species Counterpoint

Fux’s introduction of five species of counterpoint is the one still
followed today. It comprises

1. Note against note—only checks for consonant intervals

2. Two notes against note—in addition checks motion between

melodic lines

3. Four notes against note—in addition checks for dissonance

transitions
4. Ligature (Syncopation)—adds time shifting between melodic
lines
 lines

5. Florid—combines any of the first four species

Later composers would elaborate the third species so that it
could consider melodies in triple, rather than just duple, time.

12.1.4 Modal Music

Although bearing the names of the earlier Greek (Ecclesiastical)
modes, Fux uses them to refer to the starting and finishing tonic note
in his cantus firmus. They are thus the current modal scales and
comprise

1. Ionian—tonic C—corresponds to a Major scale

2. Dorian—tonic D

3. Phrygian—tonic E

4. Lydian—tonic F

5. Mixolydian—tonic G

6. Aeolian—tonic A—corresponds to a natural Minor scale

Note that Fux never uses the Locrian mode (tonic B) the darkest
of all the modes because, transposed into the key of C, it has too
many accidentals. The Locrian mode is however of great practical
use in the composition of modern jazz counterpoint.

12.2 Strict Rules Allow Freedom of

Composition
This exercise by Fux was composed and checked using Ars-Nova’s
Counterpointer software.
The example is of Fux’s third species, cantus firmus below in the
Phyrgian mode. Although there are still a few inadmissible leaps in
the melody this example shows the fluidity achievable within strict
counterpoint. Yes, you can compose dull, rote music with, or without,
rules, let your imagination guide your choices. May the force be with
you!
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_13

13. Project Code

Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

This project will allow the semi-automated composition of two part

counterpoint in the style of Palestrina or the Baroque period. It has a
number of sub-projects

1. develop a pink noise generator for the automatic creation of a

cantus firmus in a user specified ecclesiastical mode

2. develop an efficient encoding of the rule set for user specified

species of counterpoint

3. develop a list based representation of allowable voice leadings

and a user specified choice of note

4. allow user to hear, save and playback this choice and the whole
composition
5. allow user to modify consonant/dissonant groups and motion
rules to reflect a particular genre such as modern jazz
13.1 The Colours of Noise
Noise can occur in many colours, including black, but only three are
important musically

1. White describes noise with equal energy per frequency ( )

and is responsible for the high frequency background hiss of a
detuned television or radio receiver. Its main use musically is to
produce a cacophonous sound for mixing in with the synthesis
of percussion.

2. Pink noise has equal energy per octave and so is and

decays at . It is self-similar (fractal) in nature having a blend
of chaotic and correlated behaviour that makes it very useful in
music. It can be applied both to time to generate rhythmic
patterns, and to pitch to compose melodies which are tonal
rather aleatoric (random) in nature.

3. Brown noise is that exhibited in the classic random (drunkard’s)

walk where every position reached is dependent on the
immediately previous position. Its energy is and so decays
at thus making it generally too correlated for extensive use
in music. ‘Brown’ refers to the Scottish botanist Robert Brown
who observed random motion of pollen grains in water under a
microscope. This Brownian motion was eventually explained by
Albert Einstein as being due underlying molecular motion.
The functions available from importing Python’s module random
are adequate for noise generation. A very good ‘dicey’ algorithm for
generating the memory properties of pink noise was invented by
Richard Voss and described by Martin Gardner, see
Bookbackmatter. In it you roll a number of dice and sum then to get
your initial value of pink noise, then continue rolling and summing as
follows
roll die ‘one’ every throw
roll die ‘two’ every second throw
roll die ‘three’ every fourth throw
roll die ‘four’ every eighth throw
and so on with each subsequent die rolled half as often as the
previous die
The small test program ‘voss.py’ and typical output illustrates this
for eight dice,
Where the numbers represent semitone distance from the tonic over
a two octave range.

13.2 New Language Features

Project Headers given in New Language Features are applicable
only to this section.
Lists, trees as a list of lists, and list manipulation functions will be
used extensively throughout this project.
13.2.1 Project Header
You will need to use some features of NumPy and Matplotlib for this
project so enter here
import numpy as np
import matplotlib.pyplot as plt

13.2.2 Sound Storage

Arrays from NumPy will be used to hold the numerical samples of
the sound. These will be turned into WAV files for input, output and
offline processing of these sounds. Add the following line to the
project header from scipy import io

13.2.3 MIDI—Musical Instrument Digital Interface

MIDI is a wide ranging technical standard allowing instruments such
as synthesisers to be connected together within a common
framework. Notes are specified by integers in the range 0–127 with
successive numbers representing semitone intervals. Middle C (C4)
is MIDI note number 60, with C3 equalling 48 and C5 equalling 72.
The small Python program in ‘ET.py’ enables the calculation of
Equal Temperament note frequencies from a knowledge of the
reference A frequency and the number of semitones difference to the
note, e.g. concert A is 440 Hz and MIDI note 69, middle C (C4) is
MIDI note 60 and hence -9 semitones.
root=2**(1/12)
def ET(refA, semitones):
freq=refA
if semitones 0:
for st in range(abs(semitones)):
freq=freq/root
elif semitones
for st in range(semitones):
freq=freq*root
return freq
13.2.4 Composing the Cantus Firmus with Pink Noise
Generation
A minor modification to ‘voss.py’ will ensure that the pink noise
output is tonal, rather than chromatic, by specifying only the
allowable semitone spacings over a two octave range. Enter the
following Python code as ‘cantus.py’ with correct indentations

The three new features are list related

1. list declarations as a dynamic structure indexed from 0, with
[midiC3] being the tonic for starting and ending the piece
2. list elements can be summed together

3. lists can be formed by appending new elements on to the end

The only difference from ‘voss.py’ is the additional ‘while’ loop to
filter out nontonal intervals. Along with the ‘for’ loop this constitutes
the map-filter paradigm in functional programming.
Illustrated below is a typical IPython session output from the
above program where the graph’s vertical axis is the MIDI note
number.

13.2.5 Sound Generation

There are two extensions to the earlier presentation of square and
triangle wave sound presented in Chap. 9.

1. Stereo Output—this will allow the simultaneous output of the

Cantus Firmus and the counterpoint on two separate channels.

2. Continuous Frequency—this will allow notes on a tonic scale to

be output at their correct frequencies.
13.2.5.1 Stereo Output
Enter the following test program into Python saving it as ‘stereo.py’.
 The code is in four sections each headed by a highlighted
comment.
The first three sections construct three different frequencies and
durations of square wave as before. Allocation of memory for storing
all three is provided by the empty method applied to the
stereoNote array.
The final section sets the Cantus Firmus output level to 2/3, and
uses transpose to put stereoNote in the correct shape for stereo
output.
The display shows the stereo output with the counterpoint on the
upper channel, and the Cantus Firmus on the lower channel.

13.2.5.2 Continuous Frequency

This section contains two parts
1. common code applicable both to synthesis and counterpoint for
setting up the scales and multipliers

2. keeping the sampling rate constant but varying the internal

waveform data for output in stereo for counterpoint
Enter the following test program into Python saving it as
‘scale.py’.

The code is in two sections each headed by a highlighted

comment.
1. The Ionian scale in C4 in equal temperament is used with these
frequencies. The floating point frequencies are truncated to the
nearest even integer to allow for triangular waveforms. Zero
 padding is used to ensure that each note is of the same
duration..

2. This generates triangular waves as previously in Chap. 9 and

then appends zero padding.
The illustrations show the seamless joins in a scale over eight
seconds, and an enlargement of the zero padding at three seconds.

13.2.6 Sound Output of Cantus Firmus

The following Python program in ‘socf.py’ uses the list output from
‘cantus.py’, the continuous frequencies of ‘scale.py’ and a dictionary
of MIDI notes and their frequencies from ‘ET.py’ to allow playback of
a generated Cantus Firmus.
 A dictionary is an association list where key:data pairings are
separated by commas and enclosed in {}s. The first highlight shows
the creation of a MIDI note : frequency dictionary. While the second
highlight shows the extraction of frequency for a given MIDI note.
13.2.7 The Rules in Python
The rules considered are those for consonance and counterpoint
motion only at this stage.
In Python MIDI notes are used modulo 12 for checking all rule
compliances for the proposed counterpoint.

13.2.7.1 Allowable Consonances

The notes for the Cantus Firmus (CF) are subtracted (modulo 12)
from those of the Counterpoint (CP) and checked for membership of
consonance=[0,4,7,9]. In Python this is
if (CP-CF)%12 in consonance:

13.2.7.2 Allowable Motions

A list of the Cantus Firmus dynamics (rising, falling or staying same)
is compared with that of two successive counterpoint notes to
determine the motion type of Direct, Contrary or Oblique. The only
motion forbidden is Direct into a Perfect consonance (Unison,
Perfect Fifth or Octave). In Python this is (counter-cantus)$12
not in [0,7].

13.2.7.3 Allowable Dissonance Transitions

In third species only the third of four notes is allowed to be dissonant
not in [0,4,7,9] provided it is in [2,5,11] and follows in
step. This will be added in Chap. 14.

13.2.8 Output of Whole, Half and Quarter Notes for

Counterpoint
The Cantus Firmus is played in whole notes (semi-breves), whereas
the Counterpoint uses fractions of these depending on its species.

1. First Species (note against note) continues to use whole notes

(semi-breves)

2. Second Species (two note against one) uses half notes (minims)
3. Third Species (four notes against one) uses quarter notes
(crotchets)

These different note lengths will be generated as in Continuous

Frequency above with the appropriate scales for durations and
paddings. The freqSamp remains at 44100 Hz.

13.2.9 User Interface and Sound

The widgets from tkinter that will form the GUI will be

Overall a single Frame to contain remaining widgets

Overall a Button to start the program when all user entries,

including default values, are completed

Cantus two Buttons for generating and playing the Cantus

Firmus Firmus

Species a Radiobutton to select from Species 1–3 of

counterpoint

Counterpoint Buttons to record and erase, move up and down

allowable notes or remain on the same note, and
playback current notes

13.2.9.1 GUI Helper Up Down Buttons

In order to have the Up and Down buttons moving by degrees in the
modal scale use is made of dictionaries. Type the following into
‘updown.py’
up, down=1,-1
upIonian=
{0:2,1:1,2:2,3:1,4:1,5:2,6:1,7:2,8:1,9:2,10:1,11:1
}
 downIonian=
{0:1,1:1,2:2,3:1,4:2,5:1,6:1,7:2,8:1,9:2,10:1,11:2
}
def upDown(dirn, note):
note12=note%12
if dirn==up:
return note+upIonian[note12]
return note-downIonian[note12]
print(upDown(up, 64))
print(upDown(down, 67))
Note that return is unconditional so no else is required.

13.3 The Code

The project Python code is thus a composite of the above
subsections, with suitable modifications to accommodate the GUI.
The lists passed as parameters will either be as MIDI notes or
frequencies for the Cantus Firmus (CF) or Counterpoint (Cpt). Sound
output will be in stereo using square wave oscillators.

13.3.1 Project Header

Type into ‘main5.py’
from math import *
from fractions import Fraction
import numpy as np
from tkinter import *
from tkinter import ttk
freqSamp=44100
allowFreqs=
[22050,11025,7350,4410,3675,3150,2450,2205,
1575,1470,1225,1050,882,735,63
0,525,490,
450,441,350,315,294,245,225,21
0,180,175,
150,147,126,105,98,90,75,70,63
,60,50,45,42,35,30]
 allowMods=[30,25,21,18,15,14,10,9,7,6,5,3,2,1]
midiC2,midiC3,midiC4,midiC5,midiC6=36,48,60,72,8
4
midiET={36:65.41,38:73.42,40:82.41,41:87.31,
43:98.00,45:110.00,47:123.47,
48:130.81,50:146.83,52:164.81,53:174
.61,
53:196.00,57:220.00,59:246.94,
60:261.63,62:293.66,64:329.63,65:349
.23,
67:392.00,69:440.00,71:493.88,
72:523.25,74:587.33,76:659.26,77:698
.46,
79:783.99,81:880.00,83:987.77,
84:1046.50}
tonalInt=
[-12,-10,-8,-7,-5,-3,-1,0,2,4,5,7,9,11,12]
consonance,perfect, imperfect=[0,4,7,9],[0,7],
[4,9]
speciesDur={1:44100,2:22050,3:11025}
up, down=1,-1
upIonian=
{0:2,1:1,2:2,3:1,4:1,5:2,6:1,7:2,8:1,9:2,10:1,11:1
}
downIonian=
{0:1,1:1,2:2,3:1,4:2,5:1,6:1,7:2,8:1,9:2,10:1,11:2
}

13.3.2 Generating the Cantus Firmus with Pink Noise

This composes the Cantus Firmus bars in semi-breves used
throughout as a basis for the Counterpoint.

Inputs central tonic note, number of bars

Output complete midiCF

Type into ‘main5.py’
 def cantusFirmus(centreC, numBars):
def rollDie():
return randint(1,6)-biasDie
def voss(index):
listDice[0]=rollDie()
if index%2==0:listDice[1]=rollDie()
if index%4==0:listDice[2]=rollDie()
if index%8==0:listDice[3]=rollDie()
if index%16==0:listDice[4]=rollDie()
if index%32==0:listDice[5]=rollDie()
if index%64==0:listDice[6]=rollDie()
if
index%128==0:listDice[7]=rollDie()
return listDice
def pinkNoise(listPink):
pink=0
for index in range(indexRange):
pink=int(sum(voss(index)))
filtering out non-tonal intervals
while pink not in tonalInt:
pink=int(sum(voss(index)
))
listPink.append(pink+centreC)
listPink.append(centreC)
return listPink
numDice,biasDie,listDice, listPink=8,3.5,
[0,0,0,0,0,0,0,0],[centreC]
indexRange=numBars-2
listPink=pinkNoise(listPink)
midiCF
return midiCF

13.3.3 Cantus Firmus Dynamics

Generate a list from CF showing its underlying dynamics for
determination of Cpt motions.
Input midiCF
Output list of underlying dynamics
Type into ‘main5.py’
def dynamicCF(midiCF):
dynCF=[]
for index in range(len(midiCF)-1):
dir=midiCF[index+1]-midiCF[index]
if dir
dyn=’rise’
elif dir
dyn=’fall’
else:
dyn=’same’
dynCF=np.append(dynCF, dyn)
return dynCF

13.3.4 Rules of Counterpoint

Consonance, Motion and Dissonance are handled by a knowledge of
the Cantus Firmus, current note and dynamics, proposed
Counterpoint note and the Species. This returns an accept Boolean
value.

Inputs indexCF, midiCF (list), dynCF (list), midiCpt (partial list),

noteCpt (candidate)

Output accept (Boolean), list of MIDI Counterpoint notes

Type into ‘main5.py’
def rules(indexCF,midiCF,dynCF,midiCpt,
noteCpt):
def motion():
prevCpt=midiCpt[-1]
motCpt=noteCpt-prevCpt
currDynCF=dynCF[indexCF]
 if currDynCF==’rise’ and motCpt
return ’direct’
elif currDynCF==’fall’ and motCpt
return ’direct’
elif currDynCF==’same’ and
motCpt==0:
return ’direct’
else:
return ’indirect’
currInt=(noteCpt-midiCF[indexCF])%12
accept=False
if motion()==’indirect’:
if currInt in consonant:
midiCpt=np.append(midiCpt,
noteCpt)
accept=True
elif currInt in imperfect:
midiCpt=np.append(midiCpt, noteCpt)
accept=True
return (accept, midiCpt)

13.3.5 Converting MIDI to Frequency

Before being output as sound the MIDI notes must be converted to
frequencies.

Input list of MIDI note

Output list of frequencies

Type into ‘main5/py’
def midi2freq(midiList):
freqList=[]
for note in midiList:
freqList.append(midiET[note])
return freqList
13.3.6 Handling Whole, Half and Quarter Notes
Cantus Firmus and Species 1 Counterpoint are in whole (semi-
breve) notes, Species 2 Counterpoint is in half (minim) notes and
Species 3 Counterpoint in quarter (crotchet) notes. These durations
are handled by a dictionary speciesDur giving the various
durations of the sample rate freqSamp.

Inputs list of frequencies, species

Output sound data1 list

Type into ‘main5.py’
def freqData(freqList, species):
def padSaw(freqNote):
numRpts=modSamp//floor(freqNote)
sawData=np.linspace(1.0,-1.0,numRpts
, False).astype(float)
rpts=floor(freqNote)-1
sawNote=sawData
while rpts
sawNote=np.append(sawNote,
sawData)
rpts=rpts-1
padding=np.zeros(modSamp-
len(sawNote))
sawNote=np.append((sawNote,
padding))
return sawNote
modSamp=speciesDur[species]
noteData=[]
for note in freqList:
noteData=np.append(notedata,
padSaw(note))
return noteData

13.3.7 Stereo Output

Final output involves combining the wavCF and wavCpt into a two
dimensional array, list of lists, transposing this and outputting this as
a stereo ‘.wav’ file.

Inputs wavCF, wavCpt, output file name

Output None
Type into ‘main5.py’
def stereo(wavCF,wavCpt, stereoFile):
stereoNotes=np.array((wavCF, wavCpt))
stereoNotes=stereoNote.transpose()
io.write(stereoFile+’.wav’,freqSamp,
stereoNotes)
return None

13.3.8 User Interface

Control the counterpoint up/same/down key GUI Button presses to
only allow modal degree steps.

Inputs direction key, original MIDI note

Output resultant MIDI note

Type into ‘main5.py
def upDown(dirn, note):
note12=note%12
if dirn==up:
return note+upIonian[note12]
return note-downIonian[note12]
Edit the counterpoint list.

Inputs counterpoint MIDI list, MIDI note

Output counterpoint MIDI list

Type into ‘main5.py’
 def addNote(cptList, note):
cptList.append(note)
return cptList
def delNote(cptList):
del cptList[-1]
return cptList
Display the Cantus Firmus and Counterpoint.

Inputs cantus firmus and counterpoint MIDI lists, species of

counterpoint

Output None
Type into ‘main5 .py’
def display(CFList,cptList, species):
plt.plot(CFList,’bo’)
xlst,lcpt=[], len(cptList)
if species==1:
plt.plot(cptList,’r.’)
elif species==2:
idx=0
while idx lcpt/2:
xlst.append(idx)
idx=idx+0.5
plt.plot(xlst, cptList,’r.’)
else:
idx=0
while idx lcpt/4:
xlst.append(idx)
idx=idx+0.25
plt.plot(xlst, cptList,’r.’)
plt.show()
The full functionality of a Counterpoint assistant is contained in
‘main5.py’. This can be further enhanced with a full GUI as in
Chap. 14.
Part VI
On Safari
© Springer International Publishing AG 2017
Iain Gray, Snake Charming - The Musical Python, https://doi.org/10.1007/978-3-
319-60660-6_14

14. Where Next?

Iain Gray1
(1) AFIMA—Associate Fellow of the Institute of Mathematics and its
Applications, Southend-on-Sea, UK

Iain Gray
Email: iaingray@ednet.co.uk

This chapter is an ordered list of extensions and enhancements to

each project left as exercises to the reader.

14.1 Generic Header

The following in ‘header.py’ will prove useful in any extension
generic header delete those not needed
import time
from math import *
from fractions import *
from random import randint
import numpy as np
import scipy.special as spe
import scipy.io.wavfile as io
import scipy.signal as sig
import scipy.fftpack as fft
import matplotlib.pyplot as plt
import matplotlib.animation as anim
 from mpl_toolkits.mplot3d import Axes3D
from tkinter import *
from tkinter import ttk
import tkinter.colorchooser as col
freqSamp=44100
allowFreqs=
[22050,11025,7350,4410,3675,3150,2450,2205,
1575,1470,1225,1050,882,735,63
0,525,490,
450,441,350,315,294,245,225,21
0,180,175,
150,147,126,105,98,90,75,70,63
,60,50,45,42,35,30]
allowMods=[30,25,21,18,15,14,10,9,7,6,5,3,2,1]
midiC2,midiC3,midiC4,midiC5,midiC6=36,48,60,72,8
4
midiET={36:65.41,38:73.42,40:82.41,41:87.31,
43:98.00,45:110.00,47:123.47,
48:130.81,50:146.83,52:164.81,53:174
.61,
53:196.00,57:220.00,59:246.94,
60:261.63,62:293.66,64:329.63,65:349
.23,
67:392.00,69:440.00,71:493.88,
72:523.25,74:587.33,76:659.26,77:698
.46,
79:783.99,81:880.00,83:987.77,
84:1046.50}
tonalInt=
[-12,-10,-8,-7,-5,-3,-1,0,2,4,5,7,9,11,12]
consonance,perfect, imperfect=[0,4,7,9],[0,7],
[4,9]
speciesDur={1:44100,2:22050,3:11025}
up, down=1,-1
upIonian=
{0:2,1:1,2:2,3:1,4:1,5:2,6:1,7:2,8:1,9:2,10:1,11:1
}
 downIonian=
{0:1,1:1,2:2,3:1,4:2,5:1,6:1,7:2,8:1,9:2,10:1,11:2
}
14.2 Part II—Visualising Sound
Develop a fully animated circular drumhead that allows you to
specify the point to hit the drum and then to hear how it sounds. The
various stages involved are

1. animating two dimensional wave propagation using Bessel

functions

2. extending this this to three dimensional wave propagation over a

circular membrane noting that the radial components are as
specified in 1 above

3. adding user interface control over fundamental frequency and

vibration mode of drum.
Basic animation can be enhanced by

GUI adding a GUI using techniques from Chap. 9.

Sound when an enhanced percussive synthesiser is available
add drum sounds.
Animation consider the functions included in
matplotlib.animation to see if they can offer smoother
animation.
14.3 Part III—Creating Sound
The basic synthesiser can be expanded to a full synthesiser by the
following steps in order for each module
GUI complete GUI functionality as in Chap. 13.
VCO implement full frequency using the methods of Chap. 13.
VCO add pulse width modulation to the square wave oscillator by
allowing its mark:space ratio (firsthalf lastHalf) to
vary from 1:1.
VCO allow variable note length and tempo using the methods of
Chap. 13 to allow percussion synthesis.
VCO add noise generators for white and pink noise using the
methods of Chap. 13.
VCO add an all-pass filter after each oscillator to allow phase
changing. This should be done by rotating the entire ‘.wav’
data as rotating an individual waveform will introduce clicks
between waves not matching up.
VCF add full frequency input for the cutoff frequency.
VCF implement high-pass, band-pass and band-stop filters from
scipy.signal.
VCF add resonance to the filter. This is an involved problem in
Digital Signal Processing but you could replace the
scipy.signal.butter Butterworth filter with a
scipy.signal.cheby1 Chebyshev type I filter specifying
suitable ripple in the passband. As an alternative to
scipy.signal many good filter design tools. such as
FilterDesignLab-IIR shown at foot of this chapter.
EG add a second envelope generator to control the cutoff
frequency of the VCF.
EG normally the ADSR is controlled by a key pressed on and off,
simulate this with a button widget in the GUI.
EG add extra envelope stages beyond the four of ADSR.
 LFO allow targeting of the VCO to create Frequency Modulation
(FM) as vibrato, using the techniques for continuous
frequencies given in Chap. 13.
LFO use waveforms other than a sine wave for modulation.
Displays add a scipy.signal.spectogram to monitor the
harmonic content over time of the waveform.
GUI add GUI widgets to accommodate new functionality.
Duophony add two note polyphony using the techniques for
stereo sound generation given in Chap. 13.
Miscellaneous add a Sequencer for recording and playback of
melodies, using the techniques for scale generation given in
Chap. 13.

Visit https://wiki.python.org/moin/PythonInMusic to see what is

available in for instance MIDI interfacing for an external keyboard.
14.4 Part IV—Visualising Harmony
Visualisation can be enhanced by
GUI add canvas widget for drawing as in Matplotlib examples.
GUI add a beating interval to range of intervals.
Sound add playback of the interval using the techniques for
stereo sound generation given in Chap. 13.
Animation add a ‘slowed down’ animation of the displayed graphs
using the techniques given in Chap. 5.
GUI add GUI widgets to accommodate new functionality.
14.5 Part V—Composition
Basic Counterpoint can be enhanced as follows
GUI complete GUI functionality as in Chap. 13.
Cantus Firmus add Cantus Firmus above as well as below by
swapping the note octave ranges.
Modes add a complete set of modes beyond the Ionian with their
tonics as in Chap. 12.
Rules add dissonance handling and disallow leaps of more than a
perfect fifth to the consonance and motion rules already
included.
Species add species 4 (ligature) and 5 (florid). Syncopation is
achieved by rotating the Cantus Firmus list of notes.
Consonance extend allowable consonances to reflect those of
jazz scales.
Motion extend allowable motions to reflect those of jazz scales.
GUI add GUI widgets to accommodate new functionality.

FilterDesignLab-IIR
Matthias Kretschmann - author of FilterDesignLab-IIR corresponding
wedpage (http://www.mtk-digital.com/fdl/)
Curriculum Vitae
Iain Gray AFIMA Hons. BSc (Mathematics and Computational
Science) Dundee
My background is that I have recently retired, after 35 years, as a
Principal Systems Engineer working as a mathematician in radar
research for a leading avionics company (Ferranti which latterly
became Selex ES). I have over forty years programming experience
and have worked in all the major paradigms. These comprise
functional/declarative languages (Common Lisp and Prolog),
procedural languages (Fortran, C and Pascal), “combination”
languages (Scheme and Python) and object oriented languages
(Simula and Ada).
1979–81 Navigation Systems Division—compiler development
and implementation
1981–88 Display Systems Division—aircraft moving map displays
and digital mapping
1988–90 Joint Venture—development of an aircraft mission
management aid
1990–2014 Radar Systems Division—design and analysis of
tracking systems
Papers presented
1989 NATO AGARD Toulouse
“Towards a Mission Management Aid” (jointly with J Catford)
1992 NATO AGARD Edinburgh
“Planning for Air to Air Combat”
1997 Radar 97 Edinburgh
“Effects of Earth’s Curvature on Radar Tracking”
2002 GRS Radar Conference Bonn
“Advanced Tracking Systems” (jointly with Dr. B Spratt)
2004 Kalman Filtering Conference Reading
“Behavioural Analysis of Kalman Filters” (jointly with Dr. B Spratt)
Institute of Mathematics and its Applications (IMA)
1982–84 Scottish Branch Council
mid 1980s Elected Associate Fellow (AFIMA)
1986–88 Main Council
Appendix

Internet Links
These websites have been useful in preparing the document and
checking the results.
Part I

1. Python homepage https://www.python.org

2. Python downloads https://www.python.org/downloads/

3. Python documentation https://docs.python.org/

4. Tcl/Tk downloads http://www.activestate.com/activetcl/

downloads

5. Anaconda https://www.continuum.io/downloads

6. PyQt https://riverbankcomputing.com/software/pyqt/intro

7. PySide https://wiki.qt.io/PySide

8. wxPython http://www.wxpython.org

9. Python(x, y) http://python-xy.github.io

10. Audacity http://www.audacityteam.org

Part II

1. Mark Kac https://en.wikipedia.org/wiki/Mark_Kac

2. Mark Kac’s original paper http://www.maa.org/sites/default/files/
pdf/upload_library/22/Chauvenet/Kac68chv.pdf

3. Analysis of paper https://en.wikipedia.org/wiki/Hearing_the_

shape_of_a_drum

4. Drumhead modes https://en.wikipedia.org/wiki/Vibrations_of_a_

circular_membrane

5. Vibration modes http://www.acs.psu.edu/drussell/Demos/

MembraneCircle/Circle.html

6. Matplotlib tutorial https://www.labri.fr/perso/nrougier/teaching/

matplotlib/

7. scipy drum example https://docs.scipy.org/doc/scipy/reference/

tutorial/special.html

8. Matplotlib graphics examples http://matplotlib.org/examples/

index.html
Part III

1. Fourier https://en.wikipedia.org/wiki/Joseph_Fourier

2. Fourier transform https://en.wikipedia.org/wiki/Fourier_

transform

3. Bob Moog https://en.wikipedia.org/wiki/Robert_Moog

4. Subtractive synthesis https://documentation.apple.com/en/

logicstudio/instruments/index.html chapter=A

5. Moog website http://www.moogmusic.com

6. Arturia website https://www.arturia.com

7. SignalScope http://www.faberacoustical.com/apps/mac/
signalscope/

8. SignalSuite http://www.faberacoustical.com/apps/mac/
signalsuite/

9. Filter design http://www.mtk-digital.com/fdl/http://www.mtk-

digital.com/fdl/

10. Prime factorisation http://factornumber.com/?page=44100

11. Python music packages https://wiki.python.org/moin/

PythonInMusic

12. tkinter tutorial http://www.tkdocs.com

Part IV

1. Blackburn https://en.wikipedia.org/wiki/Hugh_Blackburn

2. Harmonograph https://en.wikipedia.org/wiki/Harmonograph

3. Blackburn pendulum http://paulwainwrightphotography.com/

pendulum_gallery_video_page.shtml courtesy of http://
paulwainwrightphotography.com

4. Drawing harmonic ratios http://www.fxmtech.com/harmonog.

html courtesy of Francis McConville

5. Lissajous’ curves https://en.wikipedia.org/wiki/Lissajous_curve

6. Just tuning https://en.wikipedia.org/wiki/Just_intonation
7. PrimoGraf http://leafpdx.bigcartel.com/product/primograf-
drawing-machine 0

Part V

1. Fux https://en.wikipedia.org/wiki/Johann_Joseph_Fux

2. Counterpoint https://en.wikipedia.org/wiki/Counterpoint

3. Counterpointer http://www.ars-nova.com/cp/

4. Coloured noise https://en.wikipedia.org/wiki/Colors_of_noise

5. Tuning http://www.rollingball.com/images/HT.pdf

6. MIDI https://en.wikipedia.org/wiki/MIDI
Bibliography
These books have proven useful in preparing the document and
checking the results.
Part I
M Lutz, “Python Pocket Reference”, O’Reilly, 2014
J Chan, “Python: Learn Python in One Day and Learn It Well”,
Learn Coding Fast, 2014
The above is only representative of many good Python tutorial
books
JM Stewart, “Python for Scientists”, Cambridge University
Press, 20144
A good reference on object orientated Python.
Part III
 EO Brigham, “The Fast Fourier Transform and Its Applications”
Prentice Hall, 1988
M Puckette, “The Theory and Technique of Electronic Music”,
WSPC, 2007
A Farnell, “Designing Sound”, MIT Press, 2015
A Kamenov, “Digital Signal Processing for Audio Applications”,
CreateSpace Independent Publishing Platform, 2014
R Allred, “Digital Filters for Everyone”, Creative Arts & Sciences
House, 2015
T Pinch and F Trocco, “Analog Days: The Invention and Impact
of the Moog Synthesizer”, Harvard University Press, 2004
BK Shepard, “Refining Sound: A Practical Guide to Synthesis
and Synthesizers”, Oxford University Press, 2013.
Part IV
A Ashton, “Harmonograph: A Visual Guide to the Mathematics
of Music”, Wooden Books, 2005
LP Pook, “Understanding Pendulums”, Springer, 2011
J Tyndall, “Sound: A Course of Eight Lectures Delivered at the
Royal Institution of Great Britain”, Adamant Media Corporation,
2005.
Part V
JJ Fux, “The Study of Counterpoint: From Johann Joseph Fux’s
Gradus Ad Parnassum”, Norton, 1965
D Tymoczko, “A Geometry of Music: Harmony and Counterpoint
in the Extended Common Practice”, Oxford University Press,
2011
J Collins, “Counterpoint and How to Use It in Your Music”, Echo
Pier Publishing, 2012
RF Voss and J Clarke, “1/f noise” in music: Music from “1/f
noise”, Journal of Acoustical Society of America, 63, 1918, pp
258–263
M Gardner, “Fractal Music, Hypercards and More....
Mathematical Recreations from ‘Scientific American’ magazine”,
WH Freeman, 1991
M Hewitt, “Musical Scales of the World”, The Note Tree, 2013.