Kjetil Falkenberg Hansen - Am MIT spezialisierte KI-Forschung

Kjetil Falkenberg Hansen
and Roberto Bresin
Kungliga Tekniska H ¨ogskolan (KTH
Royal Institute of Technology)
Computer Science and Communication
Lindstedsv ¨agen 24, 10044 Stockholm,
Schweden
{kjetil, roberto}@kth.se

The Skipproof Virtual
Turntable for High-Level
Control of Scratching

Skipproof is an application that emulates a typical
disc jockey (DJ) setup of turntable plus mixer and
also allows high-level control of the playing style
known as scratching. High-level control in this
case means performing with modeled, complex DJ
gestures through simplified actions: Zum Beispiel,
letting a single movement produce a sound that nor-
mally would require precisely synchronized right-
and left-hand gestures. The performer controls Skip-
proof either with the software interface or through
hardware devices connected to the computer. Der
hardware devices become alternative performance
interfaces to the standard turntable, controlling the
Skipproof application with both low-level gestures
and high-level control actions in real time. Der
mapping between hardware input and Skipproof
output is freely adaptable. Skipproof is in the proto-
type phase, but it has already been used in several
projects in recent years.

Traditionell, scratching is performed through
synchronized gestures: One hand controls the record
speed on the turntable (thus also the pitch), Und
the other hand uses the audio mixer’s crossfader to
turn the sound on or off. The crossfader is a slider
that was originally designed for fading gradually
between two turntables (or other sound sources) In
order to go seamlessly from one song to the next,
but in scratching it is instead used for turning the
sound of a single turntable rapidly on or off (often
while the other turntable is playing for instance a
rhythm track). The playing gestures are commonly
known as scratch techniques and constitute a
common language among DJs (Hansen 2002; Schmied
2006). Similarly to other instruments, traditional DJ
playing skills must be acquired through dedicated
üben, which is reflected in a growing market for

Computermusikjournal, 34:2, S. 39–50, Sommer 2010
C(cid:2) 2010 Massachusetts Institute of Technology.

teaching material (z.B., DJ Q-bert 2003; Sloly and
Frederikse 2004; Webber 2007).

Frühere Studien (Hansen 2002; Hansen and Bresin

2003) have analyzed and described DJ-performed
scratch techniques, and some of the most popular
techniques have been modeled and implemented
in Skipproof. The modeling was based on this
Analyse, with additional input gained by having an
active dialog with the scratch DJ community. Diese
models can in turn be used and manipulated in
real-time performance with simple control actions;
this makes it possible even for non-experts to play
expressively within the stylistic boundaries of DJ
playing practices.

The motivations for writing Skipproof were
to have a platform on which to model and sim-
ulate scratch techniques, as well as a tool for
studying how scratch techniques are used in ex-
pressive performances. Auch, we wanted to explore
instrument-mapping strategies for scratching and to
experiment with alternative performance interfaces
for DJs. Skipproof can be combined with hardware
and other software, and has been featured in per-
formances with DJs using quite different control
Schnittstellen (including the Radio Baton, the Reactable,
and various gesture sensors).

To the authors’ knowledge, this is the first soft-
ware that implements modeled scratch techniques
and provides high-level performance control of such
Techniken. The paper is structured as follows: Der
following section gives a background on scratching
and DJ interfaces, and then the Skipproof application
is described. Three performance situations where
Skipproof has been used are presented, einschließlich
results from informal user evaluation. Endlich, Die
current implementations and possible future uses of
Skipproof are discussed.

Skipproof is available under the terms of the
GNU Public License (GPL) and can be downloaded
from http://www.speech.kth.se/∼kjetil/software/.

Hansen and Bresin

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Figur 1. Drei
representations of the
chirp scratch.
(A) Spectrogram of two
chirps in succession
performed on the fresh
sample. (B) Common
graphical notation

(Carluccio et al. 2000) of a
chirp’s record and
crossfader movements:
The plot shows the
phonograph needle
position relative to the
fresh sample (vertical axis)
as a function of time; Die

bold part of the movement
is muted by the crossfader.
(C) Sensor output from the
record and crossfader of a
recording used in
Skipproof.

(Hansen, Fabiani, and Bresin submitted). The study
identifies how performance parameters are used ex-
pressively in scratching, and also some of the instru-
ment’s functional ranges. These and previous find-
ings have been used to set parameters in Skipproof.
DJs move the record to manipulate the playback

speed or pitch of the sound sample, but also to
vary the playing position in the sample and change
between samples. The crossfader is used to mute the
sound temporarily, either to control tone durations,
or to create short sound bursts or silent gaps
down to around 10 ms. Onset properties, or tone
attacks, are shaped both with the record and the
crossfader to make, zum Beispiel, smooth attacks of
otherwise sharp sounds. Generally for scratching,
the crossfader is adjusted so that moving the fader
nur 1 mm will change the output from muted to
maximum sound level. The fader knob can then
easily be wiggled over the small area to give rapid
onsets and offsets, generating a high tone density.
More than 100 scratch techniques have been
described (DJ Q-bert 2007), and new ones appear
regularly. The scratches are often given descriptive
Namen, such as the chirp scratch: Chirps are per-
formed in a series of rapid repetitions, bei dem die
record is moved in fast forward–backward motions
and the crossfader is switched off and on to mute
the sound at the start, at the turn, and at the end;
see Figure 1b. The onset and offset of each sound are
sharp, and the pitch quite stable and high because
of the fast vinyl movements (Figure 1a), produzieren
sounds that resemble chirping birds.

Traditional DJ Equipment

Record players were not originally intended to be
used as musical instruments. daher, they have
some less advantageous attributes, such as the
vulnerability of the needle and pick-up system, Die
vinyl as sound storage, Und, not least, the size and
weight. A very basic setup of two turntables and a
mixer weighs around 35 kg, and for a 5-hour set the
vinyl albums needed could be approximately 50 kg.
Turntables are fragile in the sense that the needle

Hintergrund

A recent study of DJ performances describes the
playing strategies and acoustics of the instrument

can easily skip and even break. To avoid needle
jumps and noise, the record should be undamaged,

Computermusikjournal

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Figur 2. Vinyl
deterioration is visible in
the ﬁve spectrograms
when comparing a record
in new condition (leftmost
spectrogram) to the state
nach 1, 2, 12, Und

15 minutes of scratching.
For this particular vinyl,
the level of wear was quite
constant between
etwa 2 Und
12 minutes. From Hansen
and Bresin (2003).

the pick-up arm correctly adjusted, and the table
and floor free from vibrations.

A vinyl record has both disadvantages and
advantages as a storage medium: Zum Beispiel, jede
side holds only around 10–20 minutes of music, Aber
offers a good visual representation of the content
and track positions by the center label, stickers,
and groove density. The physical damage the needle
inflicts on the vinyl is considerable (siehe Abbildung 2),
yet DJs endorse the worn sound.

Alternative DJ Controllers

The product catalog of commercial physical con-
trollers for DJing with other sound formats than
vinyl is growing, and academic conferences such
as New Interfaces for Musical Expression (NAME)
have reported on innovative interfaces (zum Beispiel,
Andersen 2003; Beamish, MacLean, and Fels 2004;
Hansen and Bresin 2006; Lippit 2006; Fukuchi 2007;
Pabst and Walk 2007; Villar et al. 2007; Hayafuchi
and Suzuki 2008). New physical interfaces broadly
fall into three categories: (1) interfaces converting the
rotation speeds of traditional turntables to control
Signale, (2) interfaces using the turntable metaphor,
such as with a rotating pad or jog wheel, Und (3) In-
terfaces using other metaphors, for example by
interacting with graphical representations of sound
files on touch-sensitive surfaces. Categories 1 Und 2
are currently dominant in the commercial market.

In addition to the DJ control interfaces, sequencer-
based interfaces with non-real-time interaction have
also been developed, which enable users to compose
scratch phrases, design new techniques, and play
with different samples. Zum Beispiel, Scratcher
(Faulstich 2007) is a Max/MSP patch for creating
scratch phrases taken from a list or drawn in the

TTM notation system (Turntablist Transcription
Methodik; Carluccio, Imboden, and Pirtle 2000);
and Auto-DJ (Wun, Yong, and Chan 2007) is an
application for generating mobile-phone ringtones,
using voice recordings manipulated by scratch
Techniken (chirps and stabs).

Skipproof combines features found both in real-

time control interfaces and in sequencer-based
Schnittstellen. The input devices used to control the
software can belong to any of the three aforemen-
tioned interface groups. An intuitive, easy-to-learn
interface provides real-time control and manipu-
lation of the most common scratch techniques as
performed by DJs, so that the low-level, traditional
techniques need not be practiced and mastered. Der
next section describes the Skipproof application,
including its interface and mapping.

The Skipproof Application

The name and concept came from the “skip-proof”
feature in specialized vinyl records, which was
introduced by DJ Swamp (1998). In a skip-proof
section of such a record, a sound with a duration
corresponding to one rotation is repeated for a
couple of minutes; daher, if the needle jumps out
of its current groove in the record, it will probably
stay in the same temporal location within the same
sound sample.

Skipproof was developed in Pure Data (Pd,
Puckette 1996), which facilitates control interfaces
that use common data protocols such as Open
Sound Control (OSC, Wright and Freed 1997), MIDI,
USB, or TCP/IP. Pd was chosen as it is good for fast
prototyping and easy collaborations.

A high-level scratch preset control mode and a
low-level scratch improvisation control mode can

Hansen and Bresin

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

be interchanged seamlessly during a performance.
The improvisation mode, simulating the crossfader
and turntable, is the default. The scratch preset
mode is temporarily activated when the performer
plays one of the modeled techniques.

Turntable Simulation

Skipproof provides the main functionality of a
turntable and a mixer: the user can play different
sound samples and alter the speed and amplitude
manually. The typical controllers found on the
traditional instrument—such as a start-stop button,
a pitch slider for adjusting the motor speed con-
tinuously, revolutions-per-minute (RPM) selectors,
and volume sliders—are emulated. Zum Beispiel, A
power-toggle and tap function can be used to simu-
late how DJs generate a slow, stepwise deceleration
by turning off the motor while lightly tapping the
vinyl. Ähnlich, motor-speed toggles for switching
zwischen 33 Und 45 RPM can be used to simulate a
vibrato-like effect that is created on turntables by
pressing down the RPM selectors alternately.

The volume fader, which is used sparingly in
scratching, was implemented as a linear amplitude
Kontrolle. Because the crossfader is traditionally used
in a way resembling an on–off switch, and because
scratching usually involves only one turntable,
the crossfader was implemented as a logarithmic,
triggered fading ramp going from sound to silence or
vice versa. In Skipproof, it is thus not possible to fade
between two sound sources or move the crossfader
schrittweise. Jedoch, as on a scratch mixer, Die
steepness of the fading ramp can be adjusted, Und
the crossfader direction can be reversed by setting
the end position to be muted instead of sounding (An
mixers, this is called a “hamster switch”).

Auch, mechanical properties of the turntable were
emulated and implemented as variable parameters;
these include the pitch (speed change) amount,
inertia, motor strength (torque), friction, und Hand
position on the record. Adjusting the inertia and
torque will mainly have an effect on the time
needed for the turntable platter to start up, slow
down, and restore speed. The friction parameter can
be adjusted to simulate how the friction between the

vinyl and the platter is regulated by placing slipmats
in between. The hand position parameter emulates
the effect of adjusting the hand’s distance from
the record edge (where the same gesture generates
shorter movements with a peripheral hand position
than a central).

Turntables allow one to lift and move the nee-

dle between tracks while the record is spinning
(so-called needle-dropping), and to physically mark
the vinyl with (Zum Beispiel) adhesive stickers to
quickly locate and cue the sound. Analogous fea-
tures are available in Skipproof: Needle-dropping
is possible as the sample can be changed without
affecting the playing position, and cueing is emu-
lated by a function that returns the sample to its
start.

Scratch Techniques

The main feature of Skipproof is the implementation
of twelve example scratch techniques, or “scratch
presets.” These are modeled based on the analysis of
recordings by professional DJs, and they have been
described in previous work (Hansen 2002; Hansen
and Bresin 2003). The techniques included in the
default list are baby, tear, rolltear, chop, forward,
silent back, scribble, uzi, chirp, ﬂare, crab, Und
twiddle; siehe Abbildung 3.

Skipproof has a default setting in which record
gestures and the corresponding crossfader gestures
are synchronized with each other. Jedoch, it is also
possible to decouple the two gestures for creating
new scratches and experimenting with them. Das
allows simulating a common performance practice
where new techniques are created by changing the
offset between the gestures (as can be observed in, für
Beispiel, DJ Q-bert 2007). In addition to changing
the offset between gestures, the user can create
new presets by sketching the record and crossfader
movements in tables.

We will use the previously mentioned chirp
technique as an example of how a scratch preset
has been implemented. A professional DJ recorded
isolated chirps, and chirps in sequences, verwenden
a turntable and mixer equipped with sensors to
get the rotation speed and crossfader movement.

Computermusikjournal

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Figur 3. Scratch presets in
Skipproof. The upper
diagrams in each row are
the record movement; Die
lower are the crossfader

Bewegung. In the baby,
tear, and scribble
scratches, the crossfader is
constantly on.

Tables for recorded scratches

Baby
table2

Tear
table4

RollTear
table6

Chop (short)
table8

Record –>

Crossfader –>

Record –>

Crossfader –>

Record –>

Crossfader –>

table1

table5

table7

————————————————————

Forward
table10

Silent backdraw
table12

Scribble
table14

Uzi
table16

table9

table11

table1

table15

————————————————————

Chirps
table18

Flare
table20

Crab
table22

Twiddle
table24

table17

table19

table21

table23

The best chirps from the recordings were selected
and extracted, and the sensor values stored in
lookup-tables in Pd; see Figure 1c. The tables are
resampled to have quantized durations and typical
movement ranges (based on Hansen, Fabiani, Und
Bresin submitted). A reading of the table, triggered by
the user, then sets the amplitude and deviation from
nominal playback speed, thus generating a chirp
scratch originating from the real DJ performance,
but with properties that can be changed.

Although the speed and the extent of both
the crossfader and the vinyl movements can be
manipulated in the playback of the gesture, es war
important to preserve the character of the technique:
Most techniques, like the chirp, require precise

coordination between record and crossfader in order
to sound right. Although the parameters of a preset
scratch can be changed, there is little possibility
for real-time control during the playback; a scratch
technique is typically shorter than 200 ms.

Certain scratch techniques like scribbles and
chirps are usually played in sequences of the same
Gesten, whereas others like crabs and tears can be
played singly. Chirps are therefore triggered in series
in Skipproof; Jedoch, this series can be interrupted.
As an alternative to the recorded scratch presets,
it is also possible to play idealized versions of the
scratches, which are based on equations derived
from interpolation of the gesture data from selected
Aufnahmen. These models include both record and

Hansen and Bresin

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

crossfader movements, and can be manipulated in
the same way as the recorded presets.

Audio

To simulate skip-proof sections of DJ records, alle
sound samples that are used in Skipproof have a
duration corresponding to one revolution on the
Drehscheibe (1.8 sec at 33 RPM). Technically, Jedoch,
sound files can have arbitrary length. As with a
repeated groove on a record, the sound file is looped,
so it is possible to scratch beyond the start and
ending of the sound.

For a more realistic simulation of the vinyl record
Klang, the user can choose a level of wear, achieved
by filtering the sound and adding hiss and click
noises that are looped with the sample. Deteriorated
record quality is also an important characteristic of
the scratch sound, where the high-frequency content
in the vinyl signal will be replaced by broadband
noise (see Hansen and Bresin 2003 and Figure 2).
Even audio sources other than the Skipproof
sound samples can be manipulated. Zum Beispiel, Die
DJ gestures can control sound-synthesis parameters
or—as in the case of the Reactable, explained
shortly—another system with its own audio engine.
Sound quality in digital DJ tools has so far not
been fully satisfactory; DJs particularly object to
sound quality at low speeds. Auch, even the slightest
latency between the gesture and sound is reported
as annoying. Jedoch, the main challenges for
high-level control of the models are not necessarily
latency and audio quality, but rather the controller
and parameter mapping.

GUI and Visual Feedback

Skipproof’s graphical user interface (GUI) War
designed mainly for the purposes of testing the
application and studying scratches. It was not
intended to be used as a performance interface;
Jedoch, it facilitates visual feedback when used in
combination with hardware controllers.

Each function has been made available to the
user through the GUI (siehe Abbildung 4). Zum Beispiel,

the gray rectangle in the interface represents the
vinyl record, and moving the mouse in this area will
affect the playback. In the default mapping, vertical
movement changes the speed (and direction), while
horizontal movement and position control the
hand position and friction parameters, sowie
parameters of added audio effects like echo (sehen
bottom right in Figure 4). The vertical position does
not correspond directly to the sound file position,
as this mapping was considered by DJs to be less
consistent with the idea of a rotating turntable.
Conceptually, the user moves the record with the
mouse (or other gesture controllers), and not the
needle’s position in the sample. The progress bar
immediately to the left of the waveform display
indicates the current playing position.

Compared to vinyl, digital audio makes adding
new sounds straightforward, with all music avail-
able in a searchable library. Jedoch, when using
computer interfaces, the immediate visual assess-
ment of playback position that is available when
using physical vinyl records will often be replaced
by a visual indication that is now decoupled from
the sound source and the place of the action.

Although new interfaces present new visualiza-
tion possibilities, many musicians would naturally
be reluctant to have to relearn their instrument. Als
Skipproof uses only short repeated sounds, visuell-
ization was not considered to be a critical issue.

Sensor and Parameter Mapping

Instruments intended for virtuoso playing, as most
traditional acoustic instruments are, need to be
predictable and endorse skillful handling; daher
they require low-level control. Zum Beispiel, A
violinist must carefully regulate the bow force to
achieve a desired amplitude. With computer-based
Instrumente, Jedoch, it can be beneficial to give the
player simultaneous control over several low-level
Parameter, as for instance in the case of playing a
synthesized violin on a keyboard, where key velocity
is mapped to amplitude—ranging from barely
audible to loud—without allowing imperfect tones.
The low-level control-based instrument of con-
cern here is, Natürlich, the typical DJ equipment

Computermusikjournal

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Figur 4. A screenshot of
the Skipproof GUI. Typical
instrument controllers are
placed on the left side:
motor power, Klang
Proben, volume controls,
pitch controls, and start

button. On the right side
are scratch technique
presets, preset settings, Die
mechanical parameters
such as friction, motor
properties and record wear,
and additional audio

Effekte. The middle part
has a scratch area (groß
gray rectangle), in which
mouse motions change
playback speed and
Richtung, a vertical
waveform display of the

track, Und, between these
zwei, a vertical progress bar
indicating the current
playback position relative
to the waveform display.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

for scratching, und Tisch 1 shows the mappings
between the most significant control parameters
and sound parameters for this instrument. Instru-
ment mapping for DJ scratching has been described
earlier in Hansen and Bresin (2006); and in Hansen,
Fabiani, and Bresin (submitted) it was found that
for expressive performances, durations and onsets
were varied more than (zum Beispiel) pitch. Timbre
and dynamics were found to vary mainly because of
other parameters that were being manipulated.

Control and sound parameters in Skipproof can be
mapped either like the traditional DJ equipment or
more unconventionally. Zum Beispiel, the pushing-
and-pulling gesture has become a stereotypical
trademark of DJs, but now we can explore other
metaphors for changing the playback speed—for
Beispiel, by using pressure or light intensity as the
control input.

The mapping process in Skipproof was general-
ized in Mandoux and Wohlthat (2004). Here an easy
connection and calibration of sensors was provided,

Tisch 1. Control and Sound Parameters for
Scratching with a Physical Turntable and Mixer

Duration Pitch Onset Dynamics Timbre

Record
speed

Sound

sample

Sample

Position
Crossfader
Volumen
fader

Tone

Kontrolle

•

◦

•

•
◦

•

◦

•

•
•

◦

•

◦

• = The most important features; • = Other common features;
◦ = Features that are little used or implicitly controlled.

with fast changing of the different mapping layers
between controllers and sound parameters, folgen-
ing the schemes suggested by Hunt, Wanderley,

Hansen and Bresin

Figur 5. Performance
situations with two
different interfaces for
controlling Skipproof:
(A) Radio Baton with
Skipproof GUI on a
monitor, and close-up of

ﬁnger transmitter and
antenna; (B) the Reactable
scratch objects played by
two musicians, showing
the virtual connection
between objects.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

of onsets and durations. Conceptually, breaking a
light beam could be compared to muting a sound,
but the gesture did not resemble that of moving the
crossfader. It was possible to perform fast techniques
like the crab scratch (siehe Abbildung 3, bottom) easily
by using the fingers to cross the light beam in a
comb-like way.

High-level control was used in combination with

the turntable simulation. The scratch techniques
were triggered either by foot pedals or by gestures
that approached defined areas on the antenna. Der
finger’s speed and its height at the trigger point
determined the speed and the extent, jeweils,
of the scratch.

Evaluation

The interaction was evaluated by the performing DJ
during the development phase and also after a public

and Paradis (2003). Jedoch, it is not trivial to find
sensible mappings between sensors and the high
number of control parameters available. Das ist
especially true for high-level control, where the con-
ceptual coupling is vague between (1) the simplified
control action that generates a simulated scratch
technique and (2) the combinations of gestures that
are normally needed to create the same sound using
physical equipment.

Skipproof Use Cases

In the following section, we present some use cases
in which Skipproof has been played in combination
with different interfaces and sensors. The musician’s
interaction and the resulting performances have
been informally evaluated to various extents and for
different purposes.

Scratching with the Radio Baton

The Radio Baton (Mathews 1991) is a gestural
controller developed primarily for conducting of
synthesized music sequences. A rectangular antenna
registers the 3-D position of two transmitting
batons held above its surface. The position data are
processed to control musical events.

In the first public performance featuring the
Skipproof software (see Figure 5a), the Radio Baton
was used as the turntable controller. The player wore
a transmitter fitted onto a finger, and movement
above the antenna was mapped to the various
parameters of the turntable simulation. As on
a turntable, speed was controlled by horizontal
movements on the y-axis. Vertical (z-axis) Distanz
to the antenna was mapped to hand position on the
vinyl, d.h., holding the hand close to the antenna
corresponded to grabbing closer to the record center,
resulting in more-effective hand movements. In
addition, the x-axis hand position controlled audio
effects such as echo.

In the first performance, the DJ used a standard
mixer, but in later performances, crossfading was
done by obstructing light sensed by a narrow-field
optical sensor, allowing a nimble and precise control

Computermusikjournal

Leistung. Mostly, the evaluation compared the
hardware interface to the traditional instrument. Es
was considered that having no mechanical contact
or feedback opened up new possibilities, especially
with the two added dimensions (sideways and
vertical hand position). Scratch preset playback was
reported to be a welcome feature; it was used to a
large extent during the performance.

Among the drawbacks of the Radio Baton
interface, the performer mentioned latency,
the fact that the finger-held transmitter had a
wire, movement-detection imprecision, und das
decoupling of visual feedback from the instrument.
Regarding the Skipproof functionality, the lack of
beat synchronization of the scratch techniques and
the impossibility of setting a general tempo were
indicated as problematic.

Skipproof on the Reactable

The Reactable is a circular tabletop musical instru-
ment played by controlling optically tracked objects
(Jord `a et al. 2007). The physical objects positioned
on top of the semi-translucent surface are linked
to virtual objects that connect to each other. Das
connection is visualized with a rear-projected GUI,
which in turn makes virtual connections between
the real objects. In essence, the Reactable is a
tangible performance interface for Pd.

Skipproof was integrated into the Reactable sys-
tem as a set of new or modified Reactable objects,
and it was tested with professional musicians. Der
Reactable scratch objects have been described in
Hansen and Alonso (2008). Zusammenfassend, the focus
was on high-level control with three objects repre-
senting the sample, the record movement, und das
crossfader movement, jeweils. These objects
could be manipulated, allowing the performer to
control technique type, execution properties, Und
sample position.

Unlike the situation with the Radio Baton,
the crossfader and record gestures were not syn-
chronized. This permitted new combinations and
techniques even with only a few basic movement
patterns. The objects had durations corresponding
to rhythmical beat subdivisions provided by a global

metronome. The subdivisions and durations were
controlled by the distance between connected ob-
jects and the distance from the center of the table,
in compliance with expected Reactable behavior.
To experiment with low-level control, welche

is normally not the intended operation of the
Reactable, “manual” sample and crossfader objects
were designed. The sample object approximated an
expected Reactable behavior where object rotation
mapped directly to record movement, although here
it was implemented as a temporary deviation from
a nominal speed. When the crossfader object was
moved, it would in one setting pass sound from
the sample object and in another setting mute the
sound from the sample player. This simulated the
crossfader, but deviated from the expected Reactable
behavior.

Wie erwartet, a video-based platform could not
quite perform fast enough for low-level, real-time
tasks. For high-level control, Jedoch, the latency
introduced by the image processing was not critical.

Evaluation

Skipproof on the Reactable was tested with an
expert Reactable player and an expert scratch DJ,
who both did three practice sessions followed by
questionnaire and interview evaluations (Hansen
and Alonso 2008). Between sessions, the design was
modified based on their comments. We found that
their appreciations of the interface were diverging.
Both rated high how they liked the instrument,
but overall the Reactable player got more satisfied
across sessions, whereas the DJ got less satisfied.
Already from the first meeting, the musicians
could perform rather complex and realistic-sounding
scratch patterns, and after a couple of hours of
üben, they performed expressively. They judged
that the instrument needed a lot of practice, and also
that their performance did improve with practice.

Both were satisfied with the visual feedback. Wir
found differences related to their field of expertise;
only the DJ accepted non-standard Reactable behav-
ior, whereas the Reactable expert was more forgiving
of latency and movement inaccuracy. Both found
that it got easier to perform scratching musically,
predict the instrument behavior, and have more

Hansen and Bresin

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Kontrolle, but also that it got harder to perform with
the scratch technique objects over time. One reason
for this could be that they performed increasingly
more complex phrases as they practiced, but that
the instrument could not provide a proper balance
between their control of the interface and their
musical ambition with the objects.

Scratching with a Friction Sound Model

Friction between two surfaces and scratching
a record have several analogies: The scratching
gesture is associated with the needle’s sound from
the friction against the vinyl, and the sound of
scratching is arguably less “musical” than musical
instrument sounds, but more similar to friction
noise. These observations led to an experiment in
controlling a friction sound model with Skipproof.

Serafin, Avanzini, and Rocchesso (2003) described

a physics-based model of frictional interaction
implemented in Pd. This model has many control
Parameter, such as force, velocity, and resonator
and exciter characteristics, which all need to be
carefully adjusted in correspondence to each other
to provide the intended outcome, for instance to
generate bowed string sounds.

In a recent project (Hansen, Alonso, and Dimitrov
2007), scratch techniques were used for controlling
the friction sound model of Serafin, Avanzini, Und
Rocchesso (2003). The model was controlled both
with the typical mapping where playback speed and
crossfader control frequency and amplitude, bzw-
aktiv, and with alternative mappings to control (für
Beispiel) velocity and force parameters. A combined
interface between Skipproof and the Reactable was
tested to control the friction sound model within a
specified functioning range to generate bowed string-
like sounds (Dimitrov, Alonso, and Serafin 2008).

Evaluation

The interaction was informally evaluated by the
authors and colleagues, and also by the Reactable
expert and DJ mentioned earlier. Controlling the
frequency and amplitude of friction sounds worked
well, but attempting to tune the physical model

to generate violin-like sounds with scratching was
quite problematic. Daher, even if the interaction was
found engaging, the musical output was not.

Other Examples of Using Skipproof

In addition to these examples, parts of Skipproof have
been featured in other prototypes and applications.
An ongoing project, Ljudskrapan [the Sound scraper],
is aimed at helping children with cochlear implants
and limited motor control to explore their hearing.
Among other models, scratch techniques are used
to manipulate sound samples in a playful way,
producing a more complex output than what
traditional musical instruments could.

Another interesting area of development is
Skipproof-based applications for mobile devices,
such as for creating ringtones (Wun, Yong, and Chan
2007) and for expressive interaction. Sancho (2009)
recently developed an application for superimposing
scratch techniques from Skipproof on mobile phone
music players, which lets the user trigger scratch
phrases but not perform improvised scratching.

Diskussion

Through various experiments like those described
herein, Skipproof has shown to be a versatile tool
for working with scratch techniques. Skipproof
includes only some of the basic techniques
as presets, but because most techniques are
variations on distinct record and crossfader gesture
combinations, new models can easily be added. Der
Reactable setting showed that even with a limited
subset of techniques available, the user can produce
a variety of different scratches.

Experimenting with different interfaces and map-
pings makes it possible to study how both DJs and
other players use scratch techniques expressively in
Aufführungen. With the Radio Baton, simple gestures
would trigger “perfect” scratches, but the player did
not have much control of the continuous composi-
tion of techniques, as they were triggered individu-
ally. With the Reactable, it was easier to create richer
scratch phrases as the objects would keep generating

Computermusikjournal

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

patterns while connected, and the player’s effort
could be focused on changing and manipulating
the techniques. In this sense, the Reactable offered
a more effective high-level control, which would
suggest that DJ user interfaces do not necessarily
have to resemble the traditional instrument.

When performing with sensors, we tried to
minimize the visual feedback, but when using the
Reactable, the visual feedback became an integrated
part of the instrument. Even though we did not
anticipate that visual feedback was very necessary
for short looped sounds, it turned out that the DJs
preferred such feedback as close to the source as
möglich.

Scratching is a topical case study for interface
Design, as DJs represent a large group of experts who
interact expressively with hardware intended for a
different purpose. Many scratch DJs develop am-
bidextrous skills, changing which hand controls the
crossfader and which hand controls the turntable—
this ability is imperative for certain DJ playing
styles related to scratching, such as beat-juggling.
Two-handed input is specifically mentioned as a
future research direction in human/machine inter-
Aktion (Buxton et al. 2002), and scratching can thus
be an example of an interface where both hands
independently perform highly trained actions, Und
where the users can perform the tasks with either
the left or the right hand.

Because there are many new interfaces targeting
DJs, it is worthwhile to reflect on which features
could improve the possibilities for expressing
musical ideas. A simulated turntable such as
Skipproof can, zum Beispiel, be made to go beyond the
physical constraints of the instrument. Zum Beispiel,
the scratch presets can easily produce unrealistically
fast or slow scratches—which the DJs participating
in our evaluation sessions gladly explored. Der
user can also manipulate sound sources other
than Skipproof’s internal audio samples, wie zum Beispiel
sound-synthesis parameters and external audio
engines. Jedoch, the characteristic features of the
instrument, such as physical restrictions and the
vinyl audio quality, should be regarded.

In our close contact with the DJ community we
have experienced that DJs are open to testing new
Schnittstellen, but they have high demands on low-level

control of the instrument. Although scratch DJs
may be perfectly content with existing interfaces,
the higher-level control makes scratching sounds
accessible to musicians who are eager to experiment
but less devoted to learning the instrument. Auch,
high-level control of scratching can be useful in
other musical applications such as sequencers, as is
the case for most musical instruments.

Danksagungen

The authors would like to thank the Vestax Corpora-
tion for equipment. Financial support was received
in the early stages from the EU projects Sounding Ob-
Projekte, AGNULA and S2S2 (Sound to Sense, Sense to
Sound), and recently from the EU projects Braintun-
ing (FP6-2004-NEST-PATH-028570) and SID (Sonic
Interaction Design, European Cost action IC0601).

Verweise

Andersen, T. H. 2003. “Mixxx: Towards Novel DJ

Interfaces.” Proceedings of the Conference on New
Interfaces for Musical Expression. Montreal: Kanada
Faculty of Music, McGill-Universität, S. 30–35.

Beamish, T., K. MacLean, and S. Fels. 2004. “Manipulating
Musik: Multimodal Interaction for DJs.” CHI ’04:
Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems. New York: ACM,
S. 327–334.

Buxton, W., et al. 2002. “Human Input to Computer

Systeme: Theories, Techniques and Technology.” Work-
ing draft, Available on-line at www.billbuxton.com/
inputManuscript.html. Zugriff 22 August 2009.

Carluccio, J., E. Imboden, and R. Pirtle. 2000. “Turntablist
Transcription Methodology.” Available on-line at
www.ttmethod.com/. Accessed October 2009.

Dimitrov, S., M. Alonso, and S. Serafin. 2008. “Developing
Block-Movement, Physical-Model Based Objects for
the Reactable.” Proceedings of the Conference on New
Interfaces for Musical Expression. Genova: Infomus,
Casa Paganini, S. 211–214.

DJ Q-bert. 2003. Do-It Yourself Vol. 1. ThudRumble DVD:

DIY001-DVD.

DJ Q-bert. 2007. Scratchlopedia Breaktannica: 100 Secret

Skratches. ThudRumble DVD: SECRT001-DVD.

DJ Swamp. 1998. The Skip-Proof Scratch Tool Volume 1.

LP: Decadent Records DEC002.

Hansen and Bresin

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D

F
/

3
4
2
3
9
1
8
5
5
5
0
5
/
C
Ö
M

J
.

2
0
1
0
3
4
2
3
9
P
D

B
j
G
u
e
S
T

Ö
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Faulstich, A. 2007. Scratcher. Software available on-line

Mandoux, F., and S. Wohlthat. 2004. “Design-

at www.skrasoft.com/. Accessed October 2009.

Fukuchi, K. 2007. “Multi-Track Scratch Player on a Multi-
Touch Sensing Device.” Entertainment Computing –
ICEC 2007, Lecture Notes in Computer Science. Berlin,
Heidelberg: Springer Verlag, S. 211–218.

Hansen, K. F. 2002. “The Basics of Scratching.” Journal of

New Music Research 31(4):357–365.

Hansen, K. F., and M. Alonso. 2008. “More DJ Techniques
on the Reactable.” Proceedings of the Conference
on New Interfaces for Musical Expression. Genova:
Infomus, Casa Paganini, S. 207–210.

Hansen, K. F., M. Alonso, and S. Dimitrov. 2007. “Com-
bining DJ Scratching, Tangible Interfaces and a Physics-
Based Model of Friction Sounds.” Proceedings of the
International Computer Music Conference, Bd. 2. San
Francisco, Kalifornien: International Computer Music
Association, S. 45–48.

Hansen, K. F., and R. Bresin. 2003. “Complex Gestural Au-
dio Control: The Case of Scratching.” In D. Rocchesso
and F. Fontana, Hrsg. The Sounding Object. Florence:
Mondo Estremo, S. 221–269.

Hansen, K. F., and R. Bresin. 2006. “Mapping Strategies
in DJ Scratching.” Proceedings of the Conference on
New Interfaces for Musical Expression. Paris: IRCAM,
Centre Pompidou, S. 188–191.

Hansen, K. F., M. Fabiani, and R. Bresin. Submitted.

“Analysis of the Acoustics and Playing Strategies of
Turntable Scratching.” Acta Acustica United with
Acustica.

Hayafuchi, K., and K. Suzuki. 2008. “MusicGlove:

A Wearable Musical Controller for Massive Media
Library.” Proceedings of the Conference on New
Interfaces for Musical Expression. Genova: Infomus,
Casa Paganini, S. 259–262.

Hunt, A., M. M. Wanderley, and M. Paradis. 2003.

“The Importance of Parameter Mapping in Electronic
Instrument Design.” Journal of New Music Research
32(4):429–440.

Jord `a, S., et al. 2007. “The ReacTable: Exploring the

Synergy between Live Music Performance and Tabletop
Tangible Interfaces.” Proceedings of the International
Conference on “Tangible and Embedded Interaction”
(TEI07). Baton Rouge, Louisiana: ACM, S. 139–146.

Lippit, T. M. 2006. “Turntable Music in the Digital

Era: Designing Alternative Tools for New Turntable
Expression.” Proceedings of the Conference on New
Interfaces for Musical Expression. Paris: IRCAM,
Centre Pompidou, S. 71–74.

ing of a HCI for Skipproof.” Technical report.
Stockholm: Speech, Music and Hearing, KTH.
Available on-line at www.speech.kth.se/∼kjetil/files/
Mandoux Wohlthat report.pdf.

Mathews, M. V. 1991. “The Radio Baton and Conductor
Programm, oder: Pitch, the Most Important and Least
Expressive Part of Music.” Computer Music Journal
15(4):37–46.

Pabst, A., and R. Walk. 2007. “Augmenting a Rugged
Standard DJ Turntable with a Tangible Interface for
Music Browsing and Playback Manipulation.” Third IET
International Conference on Intelligent Environments.
Ulm: Ulm University, S. 533–535.

Puckette, M. 1996. “Pure Data: Another Integrated Com-
puter Music Environment.” Proceedings of the Inter-
national Computer Music Conference. San Francisco,
Kalifornien: International Computer Music Association,
S. 269–272.

Sancho, Ö. B. 2009. “DJ Tool: A Mobile Phone Audio
Player Application.” Master’s thesis, KTH Computer
Science and Communication, Stockholm.

Serafin, S., F. Avanzini, and D. Rocchesso. 2003. “Bowed
String Simulation Using an Elasto-Plastic Friction
Model.” Proceedings of the Stockholm Music Acoustics
Conference. Stockholm: Speech, Music and Hearing,
KTH, S. 95–98.

Sloly, D., and T. Frederikse. 2004. Basic DJ Techniques.

London: Sanctuary Publishing.

Schmied, S. 2006. The Compositional Processes of UK

Hip-Hop Turntable Teams. Ph D thesis, De Montfort
Universität.

Villar, N., et al. 2007. “The Colordex DJ System: A New
Interface for Live Music Mixing.” Proceedings of the
Conference on New Interfaces for Musical Expression.
New York: ACM, S. 264–269.

Webber, S. 2007. DJ Skills: The Essential Guide to

Mixing and Scratching. Oxford: Elsevier Science and
Technology/Focal Press.

Wright, M., and A. Freed. 1997. “Open Sound Control:
a New Protocol for Communicating with Sound
Synthesizers.” In Proceedings of the International
Computermusikkonferenz. San Francisco, Kalifornien:
International Computer Music Association, S. 101–
104.

Wun, S., C.-H. Yong, and T.-E. Chan. 2007. “Musical

Extrapolation of Speech with Auto-DJ.” In Multimedia
’07: Proceedings of the 15th International Conference
on Multimedia. New York: ACM, S. 795–798.

Computermusikjournal

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
C
Ö
M

J
/

A
R
T
ich
C
e
–
P
D