Nearly a year ago I was quoted in New Scientist, commenting on some Microsoft research which
auto-creates average-sounding backing music to somebody singing. At the time I said "I suspect
musicians will argue that this is another step towards homogenised elevator music for all... But I
see a big market for this, whether it's liked by musicians or not." Click here to read that original article.
Frankly I was being polite, although my prediction seems to have come true for reasons I didn't
guess.
Since then, this research has been turned into a product called SongSmith which is getting
quite a cult following, but for all the wrong reasons... Microsoft in their wisdom created perhaps
the worst on-line advert ever. The Jury is still out on whether it's a deliberately ironic video or
whether it's just terrible. Probably the latter. But it's sorta worth watching just for the cringe-
factor. Is Microsoft really completely incapable of turning even potentially good research into a
fun and cool product..? Seems so...
I'll stick with my Mac, I think. (And ironically it appears that a Mac was even used in the video
for this PC software, although it was covered in stickers to disguise it.)
In this week's New Scientist they're doing a special issue on evolution to celebrate Darwin's 200th
birthday. They have a silly headline on the front cover, but inside they have a pull-out poster
showing evolutionary art and other research. The editor called me a few weeks ago about this and I
suggested William Latham's art, so this features heavily on the poster. They also mention some of
my old work evolving formula one racing car setups. You can download a pdf version of the poster
by
clicking here. (It's about 1Mb
in size.)
I'm still receiving lots of mail after the BBC Radio 4 programme I made. Some are a little unusual,
like the typed letter heavy on corrective fluid from a 70 year old woman who believed she had
thought of the replacement to Darwin's theory of evolution in 1973, but somehow nobody quite
recognised it... However the message below was slightly less controversial:
Dear Peter Bentley,
Unfortunately I missed your letter to Charles Darwin last week on Radio 4 and it's not available on
iPlayer!
I would love to listen to it, but in the meantime thought you may be interested to know about the
'ceramics' that I'm currently designing.
As part of my recent MPhil at the Royal College of Art I worked with a French company who have
developed some very interesting processes and materials for use with ZCorp Rapid prototyping.
I designed and made a piece called the Wedgwoodn't Tureen (see http://wedgwoodnt.blogspot.com for info) and
am now working on a group of pieces with Charles Darwin as the theme.
They will be Rapid Manufactured once I've completed adding
the texture to the pieces. I intend to pierce them with a section of AGTC genetic code.
An alternative construction method will be to use an algorithm that will build within the
constraints of the 'envelope' of the piece.
I am working on this project with Established and Sons and will be showing the completed work
at their gallery in Duke Street St. James.
I am about to place an order on Amazon for Evolutionary Design by Computers and look forward
to further engaging with the subject.
Nice to hear from you. Go to my book blog and you'll be able to listen to the radio programme
(they sent me a copy).
Your work looks great - there's certainly potential for evolving forms such as this by computer.
Good luck with it!
Many thanks for the swift response Peter. I'll go straight to
your book blog instead of my usual bedtime reading!
Looking forward to receiving your book & CD. I'll let you know of how it affects my work.
As mentioned in a
previous entry, BBC Radio 4 commissioned me to write and record a fictional "letter to
Darwin."
It was broadcast on 8 Jan, 3:45pm. They also used a fragment of my recording as
a trail for the whole series. So far the feedback seems to be positive from listeners. It's only
available on the BBC website for a limited time, so I've put the final version on my website.
You may need broadband for the file is around 7Mb.
A new UK version of the American science and tech magazine WIRED is being launched this year.
Perhaps brave in the current financial climate, but they have good publishers and their older brother
in USA has a long and excellent reputation. There'll also be a UK version of the WIRED website. One
of the new editors got in touch recently and today I met him for lunch. Apparently he found one of
my books recently and liked what he saw. So he invited me to be a contributing editor for the
magazine. Not a full time position, but it means that I may well be writing a few articles for WIRED
now and again. My other duty is to keep an eye out for interesting stories suitable for features
relating to science and technology... which I tend to do anyway.
I spent a bit of time over xmas updating my iStethoscope iphone application
(which links to The Undercover Scientist book) and adding many of the features
requested by users. As usual, for those interested in the gory details, I'll explain how it
works here.
The core features of the program remain unchanged, although I optimized the code to
make it a little more stable. Please see the entry for version 1.1 and versio
n 1.0 for details of how the audio processing works.
In version 2.0 I've added several major new features. The first and most significant is a
shake-activated phonocardiogram. This appears if you give your iphone a left-right shake
(that's a sharp movement in the x-axis accelerometer). It then replays the last 8 seconds of
audio in a loop, while displaying the wave form and animating a grey bar to show which part
is currently being played. The waveform is automatically scaled to fit, so eight seconds of
quiet sounds will show a nice wobbly waveform, but a single loud noise may cause such a
big spike that the rest of the pattern is scaled to nothing. To stop this happening when
you've been listening to your heart and then remove the device causing a loud noise, the
scaling ignores the last second of audio. So for best results when viewing the waveform
corresponding to your heartbeat, hold the mic in place for 8 or more seconds when you can
hear a clear sound, then remove and shake within one second. The phonocardiogram is
cancelled with a second shake, and recording of audio begins again. Inevitably the
resolution and sampling of the waveform is not perfect (there is only room to display 920
spikes out of 352800) so the display should be regarded as an approximation. Not all
information can be shown, so the absence of a spike on the display does not mean there is
none there. (The animation uses Layers - the waveform is written once into a
bitmap called a layer and then that bitmap is redrawn over the new grey rectangle.)
Note that there's an iphone bug which affects all apps - if the iphone goes into standby
for more than a minute while the program is running, the accelerometer is switched off. If
you find the program does not respond to movements, restart the program and it should
work again.
Another major new feature is a new mode: "Accelerometer." I added this mainly for ipod
touch users who have no microphone and so can't use the other modes of the program. In
"Accelerometer" mode, front-back vibrations (z-axis forces) are mapped to audio tones. A
small movement gives a low tone, a bigger movement gives a higher tone. Sensitivity
automatically changes according to the average movements, so several seconds of small
movements will result in the iphone becoming more sensitive. This mode is more for fun
than anything serious - there are too many interfering subsonic vibrations everywhere for
you to detect your heart reliably in this mode. Because it uses the accelerometer to make
noises, the shake-activated phonocardiogram is deactivated in this mode (the shaking
iphone symbol disappears in the corner to tell you this).
The final change to the program is the addition of help images on the first screen
(press the undercover scientist) and more hints in the pop up window about making the
most from the program. I've done another youtube video below, enjoy! If you need help,
leave a comment here and I'll post a reply.
The US version of The Undercover Scientist is different to all the others around the world.
They changed the title to Why Sh*t Happens. We'll have to see whether people appreciate
this - here's the first review of the book in USA anyway:
Why Sh*t Happens:
The Science of a Really Bad Day
Peter J. Bentley. Rodale, $16.95 (304p)
ISBN 978-1-59486-956-3
Everyone has one of those days when
nothing seems to go right, but why? Unlike
others who have broached the question,
British computer science guru Bentley
(Digital Biology) actually escorts readers
through a really bad day, exploring the
science behind all the little things that
can go wrong: he looks at why you slept
through the alarm (to explain the nature
of sleep); why you then slipped on the
spilled shampoo (a look at the nature of
cleansers and lubricants); why that torrential
downpour soaked you on your way to
work (a look at the cycle of water in nature).
This journey through the day, if
sometimes strained (getting chewing
gum stuck in one’s hair on the bus), is a
neat device for explaining the science behind
everyday things such as how clothing
is woven and why fabric is so strong
(until it rips when you bend over) and
how capsaicin in chilis fool the body and
provoke a burning sensation. Each chapter
ends with a brief tip on how to avoid
future mishaps. Hopefully, readers and librarians
won’t be put off by the title and
miss Bentley’s reader-friendly explanations
of the science behind everyday life.
(Apr.)