Sorrel Harriet

PhD Unplugged

2011-11-01T07:53:00.000-07:00

I am often asked to explain my work to those from a non-academic background, or who have little or no knowledge of acoustics. The answers I typically give are invariably either too woolly on account of my fatigue, too technical on account of my deep-rooted inferiority complex, or too dull on account of my inability to speak fluidly and enthusiastically when put on the spot. I have therefore decided to add this section to my website for interested parties to peruse at their leisure.

Why do I care whether or not the plebs understand my work? Well, because one of the ideas that motivates my PhD is my belief that people of all ages and backgrounds ought to be encouraged to engage more with the sound world. With sound being such a fundamental aspect of our emotional and physical well-being, I consider it an area in our lives that is sorely neglected. Environmental sound in particular, should be seen as a largely untapped source of pleasure and intrigue, rather than an unwelcome intrusion. I hope that my work, as well as gaining me my PhD (number 1 aim!), will ultimately prove interesting and accessible to a much broader audience than the fragment of academia it currently concerns.

An introduction to Auralisation (that word in the title of my PhD and all my publications thus far

This elusive term is increasingly finding its way into the public domain, yet without adequate explanation, such words only serve to frustrate and alienate the general public from work that should seek to engage and inspire them. Michael Vorlander writes,

“The expression ‘auralisation’ is analogous to the well-known technique of ‘visualisation’. In visual illustration of scenes, data or any other meaningful information, in movie animation and in computer graphics, we describe the process of ‘making visible’ as visualisation. In acoustics, auralisation occurs when acoustic effects, primary sound signals or means of sound reinforcement or sound transmission, are processed into an audible result. The word auralisation is used today to describe the process of signal generation, processing and reproduction as well as the result: the perceivable sound as auralisation of an acoustic problem, a room, a building, a car, or any other industrial product.”

Certainly for non-acousticians, this explanation fails to satisfy as it is difficult to see why auralisation should be ranked independently of other forms of audio synthesis. I think the main way that it differs and which should be highlighted, is the involvement of space; in short, auralisation concerns the audible rendering of sound in a space.

Thinking in terms of the visual analogy to auralisation in terms of Computer Generated Imagery (CGI), the animator seeks to place an object or actor in a 3D space. Just as an image alters according to the relationship between object and observer, so the characteristics of a sound depend, to a variable extent, on the position of the sound source(s) in relation to the listener. Using computer algorithms, the animator needs only to input a limited amount of information in order to recreate the same scene from different distances and viewing angles. Auralisation software attempts to do precisely this for sound in a space.

This is best explained using a real-world example. A live music venue wishes to predict how a performance will sound from different locations in the auditorium. This information will allow them to design their PA system accordingly. They may even go as far as to send different versions of the ‘mix’ to different speakers in the auditorium.

So how is this ‘auralisation’ performed?There are various techniques for performing auralisation. A common technique used in a lot of professional acoustic software for its speed and efficiency is known as Ray Tracing. Very simply, Ray tracing imagines the path of sound as rays fired off in multiple directions from a source. These rays are, to varying extents, reflected or absorbed by objects and surfaces in the space. Some of the rays will happen to cross paths with a receiver in a specific location. In this way, what is known as a histogram is built at the receiver by recording the passing rays and their magnitudes in time. The resulting histogram, which can be thought of as the ‘acoustic fingerprint’ of the space, is referred to in the scientific community as an ‘impulse response’: the response of a closed system (i.e. the room) to a burst of noise. Things do get a bit more complicated than that in practice, but the basic idea behind it is very simple.

Another technique, and one which I use in my own simulations, involves building a model of the system as a whole, and trying to simulate how sound waves propagate in a space. To do this, a volume of space, (or surface if one considers it a slice of zero height through a larger volume), is divided into discrete elements (i.e. chopped up into lots of little pieces of equal size). Each element represents a single point in the space. Casting your mind back to high-school physics, you may recall that audible sound is caused by the vibration of particles of air. Their movement causes tiny periodic fluctuations in air pressure which, within a certain range of frequencies, are detected by the mechanical parts of the ear and interpreted by the brain as ‘sound’. At each of the points in our computer model holds the value of the sound pressure. In the simplest of ‘wave-based’ models, we are aiming to emulate how sound propages from a point as if you had thrown a rock in a pond and wanted to observe the resulting ripples.
we are going to make the following assumptions about the behaviour of sound: Firstly, that sound is going to travel in all direction equally. Secondly, we assume that at any point in time, there is always the same amount of acoustic energy in the room – only the distribution changes. A relatively simple computer algorithm is all that is needed to update the pressure points in the space one by one at each point in time. An algorithm is simply a method consisting of a series of instructions, (often a single instruction repeated in a looping fashion), which leads to a predictable result. In this instance, all the algorithm is doing for each pressure point, is looking at the values of the point and its surrounding points in order to calculate the net energy that will remain after at the next point in time, having taken into account any losses and gains. The actual maths that is involved in the calculations is surprisingly simple for a case like this.

Ok, so now you should have a rough understanding of auralisation and acoustic simulation. If not, then I probably haven't done a very good job explaining it...so feel free to drop me an email with your suggestions on where I went wrong (either that or you are just hopelessly retarded). For now, let’s assume you got atleast 90% of it and move onto the specific area in acoustics that my research is concerned with: auralisation of sonic crystals for designing artistic noise interventions.

Sonic Crystals

Sonic crystals sound rather more mysterious than they are. In fact they are just a form of acoustic absorber based on their analogy with photonic crystals – manmade crystalline structures which scatter light in a similar way to natural crystals. The lattice structure of crystals that is responsible for their intense colouration or opalescence. What is essentially happening inside a crystal lattice is that waves are being bounced around inside a structure containing parallel surfaces. At certain frequencies, where some multiple of the wavelength or half wavelength matches the distance between the surfaces, there will be some cancellation or ‘summing’ of that particular frequency component. The end result, sticking with the example of natural crystals and light waves, is that certain frequencies of light are absorbed completely by the structure, while others are reflected with a greater intensity.

So, basically, what we’re dealing with here is a very selective kind of absorber. Now, in an environmental noise context, one could conceive of wishing to attenuate a particular range of frequencies whilst leaving others in tact, in order to filter or ‘shape’ the noise into something more interesting and more pleasant. For example, the sound of a river is widely considered relaxing, partly due to its association with nature, but also due to its frequency distribution. Conversely, a busy road is disliked by many for the same reasons. Perhaps, using a carefully designed ‘sonic crystal’ as a filter, it would be possible to make the road noise resemble the sound of a river. Now obviously there are other reasons why a river is preferred to road noise – for example, the temporal structure of the noise is a major factor - but frequency distribution is by no means a trivial aspect of the larger problem. Even if it were to be addressed in isolation, it might have a significant effect on how people perceive and react to a particular noisy environment.

To be continued...

Still to come

What sounds do people prefer, how do we know that, and why?

How might sonic crystals help?

Let’s see some examples then...

New pencils

2011-09-01T00:58:00.000-07:00

Tried out my new pencils :-) And a slightly different style too.

Ullswater

2011-08-29T03:16:00.000-07:00

Not booking ahead has its advantages, although, the woman in the campsite took a bit of convincing that, actually, we'd much prefer to be up the hill in the overflow campsite, miles away from the campsite conveniences and swathes of fellow campers. As it happened, things turned out just as we'd envisaged...a broken night's sleep on a blustery hilltop after a veritable feast of baked beans and chocolate.

Holiday!! :-D

Temporal smearing possible??

2011-08-26T08:02:00.000-07:00

Of greater significance than the frequency content of noise, is the temporal structure, with frequent abrupt changes in the signal amplitude resulting in a more annoying sound than one of more or less constant amplitude. As well as altering the frequency content, there ought to be some way of `smearing' the signal in the time-domain...I guess this amounts to reducing the clarity of the signal (i.e. altering the ratio of direct to reverberant sound). I was wondering about using Helmholtz resonators to achieve this...just a vague thought anyhow. Will come back to it.

The Misanthropist, or, Yet Another Miserable Fucker at War with Himself

2011-08-21T09:56:00.001-07:00

Radioactive Orchestra

2011-08-12T06:29:00.000-07:00

Came across the Radioactive Orchestra while browsing the New Scientist website. Pretty cool, and a slick way of illustrating the curious and startling relationship between music, maths and the natural world :-)

London 2011

2011-08-11T13:20:00.000-07:00

This summer I took a little jolly to London where I felt temporarily revitalised and borderline-enthusiastic about life before the familiar state of boredom and disillusionment returned.

Wake me up when it's all over

2011-08-11T12:28:00.000-07:00

Here we have a perfect example of why I hate this fucking country and must leave as soon as I'm done with the whole Pissing Hours down the Drain malarky. To be fair, it's not the country itself I dislike so intensely, it's the fucking idiots who are relentlessly sapping whatever vestiges of hope I had left for humanity. There's just too many of them in too small an area to make this a viable resting place.

For the record, this guy (I'm guessing it was a guy), had parked outside my house leaving me no choice but to back up as close to his front bumper as possible in order to avoid blocking my neighbour in. I *did* make sure he had plenty of room to reverse, but the fact that the thought had crossed my mind about how he was going to get out apparently never entered his.

Minimal Surfaces

2011-08-08T13:51:00.000-07:00

More research on gyroids and minimal surfaces, came across these artists on Wikipedia. Pretty cool.

Continuum (1979) by Charles Owen Perry

Arabesque XXXI carved in bubinga wood by Robert Longhurst, c. 2007

Gyroid Sonic Crystals?

2011-04-04T07:14:00.000-07:00

So the theory goes, 2D phononic crystals are less effective 'selective' absorbers than 3D phononic crstals. A 3D phononic crystal will typically be constructed of ball bearings suspended in a resin medium. To qualify as a sonic crystal, the supporting medium must be a liquid (i.e. air), but trying to support a periodically spaced array of ball bearings in air is obviously going to cause problems... Two articles I came across recently have prompted me to start toying with the idea of gyroid lattice-type sonic crystals. The first was an article about the photonic reflectivity of natural gyroids or `biophotonic crystals' as they are also known (National Geographic, 2010). For example, the type identified on a particular species of butterfly's wing scales. These have also been evaluated in FDTD simulations (Michielson, Journ. of the Royal Soc., 2009). The second article was about 3D printing, proclaimed by some to be the next revolution in manufacturing technology (Economist, 2010). I wondered whether, using 3D printing technology, it might be possible to manufacture gyroids on the scale of sound waves that would be effective selective absorbers. The material would still have to be a hightly reflective - i.e. the oil based polymers typically used in 3D printing probably wouldn't be effective.But apparently 3D printing in stainless steel is also possible...

My first gyroid in Matlab

Icon conundrum solved

2011-01-26T14:17:00.000-08:00

Essential to the development of any software are the icons of course. Thanks to this fantastically useful little site: http://www.favicon.cc I have been able to save myself heaps of phaffing about in Fireworks.

Subversion Edge interferes with Python

2011-01-13T04:27:00.000-08:00

Installed subversion edge and tortoise svn. Subversion edge by Collabnet is a free svn host. However, after installing, my Python IDEs would not start. Uninstalled Svn Edge and working again fine now. Think it changes the PYTHONPATH and PATH Python variables, but have not tried to find a workaround yet. May do at a later stage. Must remember Subversion Edge runs as a system service, so it only sees the system-wide environment variables - might be the key to finding a solution.

Building and Installing VTK with Python 2.7 in Windows XP

2011-01-10T07:37:00.000-08:00

Hope this may be useful to someone somewhere...it certainly took me a while to figure it out. It should be much more straightforward in Ubuntu, but if XP is unavoidable, try this:

Download and install the latest version of Python (www.python.org/download/)

Download and install a free build system such as cmake (www.cmake.org/cmake/resources/software.html

Download the VTK source files (www.vtk.org/VTK/resources/software.html)

Download and install activetcl (www.activestate.com/activetcl/downloads). Skip this step if tcl/tk already installed on system.

If necessary, download and install a C++ development environment such as MS Visual C++ Express (www.microsoft.com/express/Downloads/#2010-Visual-CPP).

Unpack the VTK source somewhere.

Make a subdirectory named "build".

Open ccmake.

Where it says `Where is the source code', enter the path to the main VTK directory.

Where it says `Where to build the binaries', enter the path to the new build folder you just created.

Hit `Configure'.

Toggle BUILD_SHARED_LIBS on.

Keep VTK_USE_CARBON on.

Do not use VTK_USE_COCOA.

Toggle VTK_USE_HYBRID on.

Toggle VTK_WRAP_PYTHON on.

Configure again. If there are errors, they are most likely requiring you to add a couple of missing entries to the list of variables. You will probably need to add PYTHON_LIBRARY (eg. c:/python27/Lib) and PYTHON_INCLUDE_DIR (eg. c:/python27/include).

When no errors are returned, hit `Generate'.

Find the file FindPythonLibs.cmake and check it has the version of python you are using in the list of version numbers.

From the VTK build folder, locate the .sln file and open in VC++.

In VC++, set the debug option to `Release' or `ReleaseWithDebug'.

Build the solution.

Now you should be able to go into your Python IDE (I use Spyder* which is free), or Python shell, and do `from vtk import *;' to start using the VTK library. Simple as that! I will add tutorials as I learn myself, or if you have any tutorials, please send them my way!

*You must have SciPy and NumPy packages installed too.

Helsinki

2010-07-10T08:04:00.000-07:00

Sketches I made while on holiday in and around Helsinki. Actually, some of them may have been made while daydreaming.

Aglaia at the top of Rosedale Chimney

2010-05-29T10:41:00.000-07:00

April 2010

2010-05-07T07:16:00.001-07:00

Tortoise

2010-02-06T14:37:00.000-08:00

Portrait of a young woman

2010-02-05T13:32:00.000-08:00

Sketch of a girl

2010-01-10T13:34:00.000-08:00

A Serious Man

2010-01-02T13:12:00.000-08:00

Tribute to one of the greatest portrait artists of the modern age. He's dead now of course.

Westmacott bust Ashmolean

2009-12-27T18:58:00.000-08:00

Our Father

2009-12-27T13:36:00.000-08:00

Portrait of a man in a blue t-shirt

2009-12-17T13:27:00.000-08:00

Frog

2009-12-01T11:52:00.000-08:00

Dogs

2009-05-29T10:43:00.000-07:00

Noah and Marl. Marl would not sit still. She is always fidgeting.