Visualization of Music at Arm’s Length

João Tiago Gomes, Maria Beatriz Carmo and Ana Paula Cláudio

LabMAg, Faculdade de Ciências, Universidade de Lisboa, Lisboa, Portugal

Keywords: Natural Interaction, Leap Motion, Visualization, Musical Composition.

Abstract: This paper presents an interactive application to compose music. It aims to give anyone the possibility of

creating music from various sounds that can be composed on top of each other to create music of increased

complexity. The musical composition is accompanied by a 3D visualization that is intended to serve not

only as a guide to the user’s composing process but also as a visualization of the music being created. A

user study was conducted to assess the composition process, the interaction with a Leap Motion device and

the correspondence between sound and visualization.

1 INTRODUCTION

Lev Sergeevich Termen patented in 1928 an

electronic musical instrument called Theremin. The

musician controlled this instrument without any

actual physical contact. The instrument has two

antennas that control the pitch (or frequency) and the

amplitude (or volume) of the sound. Each antenna is

controlled with one hand and the musician brings his

hands closer to the antennas in order to control the

sound the instrument produces.

In 2012 the Leap Motion device was announced.

This device tracks the position of the user’s fingers,

hands and forearms very accurately (1/100 of a

millimetre) in a semi hemispheric space of about one

meter. The Leap Motion device made it possible to

mimic the interaction with a Theremin.

Besides playing an instrument, several musicians

create music in real time assembling sounds

produced wheter by voice or by instruments. Several

contemporary musicians have been following this

approach for some time. This is only possible using

recent technology (hardware and/or software). This

trend is not specific to a musical style. Nowadays

there are artists from several musical styles that use

this process to produce music.

The goal of this work was to develop an

application with three components: first of all, a

musical component that would enable the user to

play several sounds and compose them as he wishes;

in second place, to use the Leap Motion device to

interact and create music, trying to keep this

interaction as natural as possible; and finally, to

visualize music by clearly identifying the various

sounds used in the composition. These three

modules, although separate, have to work together in

order to provide a good user experience. This is even

more relevant with a new form of interation that is

not the keyboard and mouse. By using the Leap

Motion device we have to take into account new

problems and challenges that do not exist in

traditional interfaces that have been studied for a

long time.

After presenting related work in section 2, the

components of the application are explained in

section 3. The results of a user study are presented in

section 4 and in section 5 are shared the main

conclusions and future work is pointed out.

2 RELATED WORK

In this section we focus on projects developed in the

musical area to compose music with an incremental

process and on web applications for music

visualization.

Musical Process

The work of several contemporary artists was

analysed (DubFX, Zöe Keating, Tom Thum e Jarle

Bernhoft). These artists produce music all by

themselves by playing and combining several

instruments and sounds. First of all the artist has at

his disposal pieces of equipment (hardware) and/or

software that allows him to record several tracks.

The artist starts by recording a single track, whether

his own voice or an instrument. After this, the sound

454

Tiago Gomes J., Beatriz Carmo M. and Paula Cláudio A..

Visualization of Music at Arm’s Length.

DOI: 10.5220/0005312604540461

In Proceedings of the 10th International Conference on Computer Graphics Theory and Applications (GRAPP-2015), pages 454-461

ISBN: 978-989-758-087-1

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

that was just recorded starts playing immediately in

an infinite loop. Next, the artist records another

sound in a different track, while the previous one is

playing. This process is repeated several times in

order to overlap the several tracks to create more

complex music.

Real Time Music Visualization

Several applications have the ability to present

music visualization in real-time. They are able to use

information about the sound frequency at a given

moment and use that information to create the

visualization in real time. The most common

example of this is a simple bar equalizer that is

present in some musical devices, such as stereos.

Some applications only possess a single 3D

visualization, like Loop Waveform Visualizer (Loop,

2012), Music Colour Particles (Color Particles), and

Cube Visualizer (Cube Visualizer), whereas others

have several. These more simple applications make

use of already known musical visualizations, for

example a sound wave, and transform it in a 3D

version. While A Dive in Music (Dive) has more

than 20 visualizations, all very complex and

involving always particle emitters.

Pre Processed Music Visualization

The clear advantage of pre-processed music

visualization when compared to real time

visualizations is that the end result can be much

more rich in graphic terms because much more

processing power can be used to generate such

visualizations. This processing power can be

devoted purely to image rendering or distributed also

to sound analysis. For example, when comparing the

previous example A Dive in Music with Ljósið

(Ljósið), it is clear the difference between the

particle quality in each example and in the overall

graphic impact that this quality causes on the user.

Another example of pre-processed applications is

one from Christopher Garcia (Garcia, 2011). This

author uses a two-step process to create a musical

visualization. First, he analyses the music and writes

information to a file that is then used to serve as

input to create the visualization itself.

Visualization and Creation of Music

Some applications offer the user the possibility of

creating music and visualising it at the same. In this

case, the applications are mainly focused on the

creation instead of the visualization, which means

that the interactivity plays a key role.

Several applications were analysed, either with

3D or 2D visualizations, but they were all very

similar: ToneMatrix (ToneMatrix), Beat Petite

(BeatPetite) and ToneCraft (Tone Craft). In these

applications it is possible to play different sounds at

different moments in time. The metronome that

counts the time and regulates the moments when

notes are played has a significant role in these

applications. When the bar comes to an end it

returns to the beginning and therefore loops

infinitely. In each of the applications it is possible to

play a total of sixteen notes in each bar (wherein the

application’s metronome is in an infinite loop).

The ToneMatrix application, besides using a

metronome and repetition as stated before, it uses a

slight delay between beats. This delay coupled with

the repetition of the metronome gives music

produced by this application a jazz tempo.

3 VISUALIZATION OF MUSIC AT

ARM’S LENGTH

We developed an application to create music

combining predefined sounds. This application has

three main components: a) audio, that manages

playing and recording various sounds and provides a

mechanism to keep track of time in a precise way; b)

user interaction, both by using the Leap Motion

device to play and compose the various sounds and

through a graphical interface to also compose the

several sounds; c) visualization of the created music.

This was accomplished using the javascript

language, HTML5, CSS3 and Web Audio API and

was built for the Google Chrome browser. The final

application can be available on the web, not

requiring any additional download: the user only

needs to own a Leap Motion device.

3.1 Audio

The audio component was the most challenging. It

has two different roles. One is the ability to count

time in an exact and reliable way. The other one is to

make available several sounds to be played, organize

several tracks and be able to handle the process of

playing and recording each track.

Taking into account the technology used to

develop this application, it was difficult to

accomplish all these requirements and avoid any

major delays in terms of sound synchronization and

performance.

3.1.1 Audio Graph

The web audio component was made using the Web

Audio API whose main paradigm to handle audio

consists of a graph of several linked nodes that route

VisualizationofMusicatArm'sLength

455

the audio from its sources to its destination, as

shown in Fig. 1.

The entire graph is constructed in the

initialization moment. This means that all the nodes

are created and their relationships are established.

After this initialization phase, several properties of

each node are manipulated, for example, changing

the sound level of a track.

The graph can be divided in two groups: the

nodes that are common to every track in the

application (right side of the graph) and the tracks

themselves. Regarding the tracks, we considered two

types: the base tracks and the normal tracks.

The common section is composed of four nodes:

a Gain Node, named Master Volume, which controls

the overall sound of the application; a Dynamics

Compressor that prevents sound distortion and to

achieve this it boosts lower sounds and decreases

higher ones; an Analyser Node, called Master

Analyser, to analyse the sound that passes through it

and pass that information up to the developer. The

Destination Node, located at the right end of the

graph, is the one that eventually every other node is

going to connect to, whether directly or indirectly.

This node represents the computer’s audio exit,

usually the speakers.

Figure 1: Audio routing graph.

There are two base tracks: one has a sound of a

drum base and the other one has a synth sound that

produces a continuous and surrounding sound. The

base track that possesses the base drum sound has

three nodes: a Gain Node, called Volume, which

controls its volume; an Analyser Node, simply

called Analyser, which sends information about this

track’s frequency levels to be used to generate the

visualization of this track; and a Source Node that

emits the sound.

The remaining base track that contains the synth

sound is more complex. It was adapted from an

existing library. It produces sound by using up to 20

random sound generators. Each one of these random

sound generators uses a Script Node Processor to

generate sound in a random fashion. The sound

passes through a Biquad Filter Node that is

configured to the BANDPASS mode and limits the

range of frequencies that pass through this filter.

Finally a Panner Node is used to make the sound

that passes through it sound like its coming from a

different direction every 0.5 seconds. This produces

a random sensation in the user.

Each one of these random sound generators has

a specific frequency that serves as input to the

Biquad Filter Node, so each generator plays a

specific frequency that is randomly selected from a

restricted set of frequencies. This makes that final

sound is enjoyable.

Finally, the normal tracks are also composed by

three nodes, as the base drum base track is. There

are currently five normal tracks but can easily be

added more.

3.1.2 Metronome, Marking Time and

Performance

Marking time correctly and building an exact

metronome was crucial. Performance was therefore

a critical part of the application, keep in mind that

the application’s visualization, audio component and

interaction with the Leap Motion device are all in

real-time. Taking into account that the application

runs in a web browser, it has a handicap because of

the browser’s clock precision.

Based on the number of BPM (Beats Per Minute)

and of notes that the user can play in each bar, a

millisecond interval is defined. This is the interval of

time that separates each beat; each time this interval

of time has passed the metronome “awakens” and

knows that has to perform a series of actions; those

actions in javascript translate into a callback

function. This function is responsible for: start

recording a track; end recording a track; update the

counter that is responsible for knowing when to stop

recording a track; play sound if the user wants to

play the sound of a specific track; play the sounds of

the base tracks; in case there is a track with a

recorded sound and if it is time to play it, in fact play

it; lastly, get the correct counting of notes.

As was mentioned before, there is a slight delay

of a few milliseconds in the browser’s clock. This

makes the metronome “awakening” not the same

throughout time. Also, the cumulative effect of all

GRAPP2015-InternationalConferenceonComputerGraphicsTheoryandApplications

456

individual delays causes even more problems on

terms of performance.

A strategy to solve this problem was to schedule

the next call of the next “awakening” not to the

expected value but to adjust it so that it takes into

account the previous delay. Thus, the cumulative

error of each callback is eliminated, remaining only

the error of single calls.

This technique reduces to acceptable levels the

clock delay between callback calls. However, inside

the callback function, the operations must be

executed in sequence to avoid audio discrepancies.

Therefore a critical zone was created were

operations are executed consecutively. These

operations are prepared previously.

Threads were tested in this audio component but

they proved inefficient because of the

communication delay between the main thread and

the secondary thread.

3.1.3 Tracks

The application has several tracks and each one

plays a specific sound. When the application

initializes, the various sounds are loaded from .wav

files. Each sound is of a single instrument or part of

instrument (drum sounds for the most part). As

stated before there are two types of tracks, base

tracks and normal tracks.

The difference between them is the level of

control the user has over them. The base tracks are

played automatically from the start of the application

to the end. The normal tracks are controlled by the

user, specifically through a Leap Motion device.

Each track sound was chosen taking into account

how it sounded by itself and together with other

sounds, but most importantly because of its duration.

As has been explained before, there is a time interval

between callback calls and the duration of each

sound cannot exceed that interval because sounds

cannot overlap.

3.1.4 Recording Sound

Several approaches were tested to record sound. The

first one was to record sound using an existing

library that uses javascript nodes and threads to

accomplish that task. However, the result did not

have the needed precision to start recording and end

recording at an exact moment, because of the latency

that javascript nodes and threads have when used

with the Web Audio API.

Another solution was used: record the moment

when the sound was played instead of the sound

itself. That is, when the application is in recording

mode, it knows that is recording a given track and

knows when a sound is played. Hence, if a sound is

played, it is stored in a boolean vector as true, if a

sound is not played then the value is set as false.

3.2 Interaction

The user’s interaction with the application is made

in two complementary ways: through the use of a

Leap Motion device to play sounds and by using the

graphical interface to control the several tracks.

3.2.1 Interaction with the Leap Motion

The user interaction with the application regarding

music composition is made through a Leap Motion

device. The user is capable of controlling rhythm

sounds, meaning that these sounds do not prolong in

time: they are played and shortly after turn to

silence. A counterexample is a violin sound where

the player can prolong the sound for some time. As

the rhythm sounds end rapidly it was decided to give

the user the ability to control two parameters: first,

the moment when the sound is played (this moment

can be one of sixteen pre-established moments that

can be played) and also the volume of the sound.

Once a track is selected and therefore a sound is

also selected, a user can play that sound. The user’s

left hand controls the moment the sound is played.

Optionally, the user can use his right hand to control

the volume of the sound that is being played. To

identify hand positions boundary planes were

defined, both horizontal and vertical. So, a vertical

plane parallel to the yy and zz axis was placed at the

middle of the scene and it separates the zone of

action of each hand (Fig. 2).

For the left hand were defined 4 vertical zones,

each one corresponding to a specific action: the

higher one prevents any sound from being played

(z1); the next zone plays the sound at a normal rate

(the same rate of the metronome) (z2); below this

one, sound is played at double rate (twice the speed

of the metronome) (z3); and finally, the bottom zone

Figure2: Screenshot of the application where the different

action zones are visible.

VisualizationofMusicatArm'sLength

457

plays the sound at a rate four times the metronome

(z4). The speed at which the sound is played is

directly related with the speed of the metronome as

well as the number of notes that can be played and

the moments at which they can be played also.

In table 1 the cells noted with "x" correspond to

the moments when sound is reproduced according

with the zone where the left hand is.

The right hand of the user controls the volume of

the sound that is being played. For this hand,

boundary planes were also defined to separate

different zones of action: the higher zone matches

the higher volume (80%), the lower zone matches

the quieter volume (10%) and in between a

proportion was made between the position of the

user’s hand and the volume itself. The extreme

zones occupy both 25% of the space available and

the intermediate zone occupies the remaining 50%.

Fig. 2 shows the virtual hands and the boundary

planes. As no track is selected, the boundary planes

are coloured with a neutral colour as well as the

virtual hands. When a track is selected, both the

virtual hands and the boundary planes change colour

to give feedback about what is happening on screen.

3.2.2 Graphical Interface to Control Music

and Track Time

The graphical interface (Fig. 3) is composed of three

main areas: the track zone, the master volume zone

and the time zone.

In the lower right side of the screen there is a

slider labelled as Master Volume that controls the

volume of the entire application, that is, the volume

of the Master Volume node in the audio graph.

In the bottom centre of the screen are 5

rectangular areas that represent each one of the

normal tracks. By selecting one track the user can

play the sound associated with that track. Each track

area has two buttons: one to record and one to either

pause or play, depending on the current state of that

track.

By pressing the record button the top half of the

track area turns orange to indicate that the

application is preparing to record (the actual

recording happens at the beginning of the bar, the

user can guide himself by the numbers that keep

track of time that will be explained ahead). When

the track actually begins to record, the area turns red.

Once the recording is finished, it turns green to

indicate that the track is now playing in loop. Then

the user can choose to either pause that track, and

the track area turns brown, or record again erasing

the previous recording.

Finally, in the top centre of the screen there are

two numbers in the X/X format. The left one counts

the tempo of the metronome, from one to four, and

is synchronized with it. The second number, the one

on the right goes from one to sixteen and represents

the number of notes the user can play with his left

hand when his hand is in the zone z4 (table 1). The

second number is related with the first because it

divides each tempo in four. This means that when

the first number has the value one, the second

number has the value one and then becomes two,

then three and finally four, as the second number

turns to five, the first becomes two and so on.

3.3 Visualization

One of the goals of this application was to match the

sound heard by the user with the visualization of that

sound. As mentioned before, the application has two

types of tracks: base and normal tracks. To visualize

the base track that reproduces the sound of a drum, a

vertical plan was placed at the back of the 3D scene

containing two circles, positioned in the left and

right lower corners of the plan (Fig.3). These circles

grow in size as they react to the sound of the base

track and resemble a pulsing light. This visualization

is initiated as soon as the application begins.

The normal tracks can be manipulated by the

user. To each track corresponds a block, composed

of several parallelepiped. All the blocks are aligned

horizontally (Fig. 3). These blocks are animated and

move when the corresponding track plays sound. To

identify if a track is playing, three types of

animations were defined: one to the extreme blocks,

one to the middle block and one to the second and

fourth blocks (Fig. 4).

Table 1: For each action zone is displayed the moment when the sound is reproduced.

xx xx

x x x x x x x x

x x x x x xxxx x x x x x x x

Zone vs

time

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

GRAPP2015-InternationalConferenceonComputerGraphicsTheoryandApplications

458

Figure 3: The normal tracks are not playing.

Figure 4: The normal tracks are playing.

The blocks on the extremes left and right are

composed of 20 parallelepipeds with the largest

dimension parallel to the zz axis and perform a scale

transformation along the same axis. The middle

block consists on a set of 5x5 parallelepipeds and

performs a scale transformation in the yy axis.

Finally, the remaining two sets are also composed of

20 parallelepipeds with the largest dimension

parallel to the zz axis and perform a translation along

this axis in an uneven way, in other words, if the

first parallelepiped translates positively in the zz

axis, then the second parallelepiped translates

negatively, and so forth. The animations of the

blocks use a random factor to affect their size.

4 USER STUDY

4.1 Methodology

A user study was performed involving two sets of

users: experts and common users. The experts were

divided in two groups, music experts and interaction

experts. For each participant, an interview was

conducted in the following way: first of all a brief

explanation of how the application worked was

given, after that the participant experienced the

application freely and finally he answered questions

posed by the interviewer. The questions aimed to get

feedback about the usability and quality of the

application but also to gather opinions about future

improvements that can be made.

The questions had three different formats:

• Rating questions where a one to five scale was

used and the user was asked to classify a feature

of the application.

• Choice questions where two alternate answers

were presented and the user was asked to choose

one of them.

• Semi-open choice questions where the user was

asked to choose an alternative and was invited to

suggest alternatives.

The interviews took in average 30 minutes with

common users and about 1 hour with experts. The

set of questions was the same, but the open choice

questions were answered with different levels of

detail.

4.2 User Profile

The tests were performed on a universe of 23 users

aged between 17 and 50 (average 28). In order to

define each user profile, it was asked the user’s age

and gender, if he usually plays computer/console

games; if he knows how to play an instrument; and

finally, his dominant hand (right hand, left hand or

both). 78% had some kind of experience in playing

with computer/console and 77% of the participants

knew how to play at least one instrument.

Due to their background and training some of the

participants were considered experts in the musical

or interaction areas. The musical experts know how

to play an instrument and have also in depth musical

knowledge or are professional musicians or know

how to work in musical production. The interaction

experts, because of their job or academic career,

handle and build user interfaces on a regular basis.

Four people were considered musical experts and

two were considered interaction experts.

4.3 Result Analysis

Not all interviews were performed in a computer

that offered the best user experience. Taking this

into account, a rating was given to each computer

where each interview was performed and the results

were analysed with this in mind. However, this was

not as problematic as was expected, firstly because

only four interviews were conducted in a poor

computer; secondly because even in these cases

there was only a reduced number of questions that

escaped the norm of answers. Finally, the answers

that escaped the norm are duly referred in the

analysis.

VisualizationofMusicatArm'sLength

459

Audio

The interviewed users were happy with the choice of

rhythm sounds that were present in the application

(100% greater than or equal to rating 4 and 52%

with rating 5). When asked if they would like to

experiment sounds that were not only rhythmic, the

result was unanimous, 100% said yes.

Beyond sounds that could be controlled by the

user questions were made regarding the base track.

The assumption at first was that it would help the

composition process for someone that was not very

experienced in an application of this type. This

assumption was confirmed by the answers given

with 78% saying that the beat sound always present

was a plus because it gives some sort of foundation

to work upon. 18% of the answers were in the

opposite way and a user suggested that this should

be something that the user could turn on and off at

will. It is worth mentioning that from the negative

responses (only four), two were given by experts.

This is important because experts tend to be more

creative when using the application and do not need

this artificial “crutch”. The next question was about

the beat speed, specifically BPM speed (it was set to

90 BPM, i.e., beats per minute). For this answer the

scale was: 1 – speed much slower, 2 – speed slower,

3 – ideal speed, 4 – faster speed, 5 – much faster

speed. So the assumption here was that the results

would be expected closer to the number 3 instead of

5. This assumption was confirmed with 75% of users

giving rating 3 to this question. The remaining gave

an answer that was not in the questionnaire, they

answered the speed should be adjustable. Two of

these answers were given by experts.

Finally, the last feature tested was the recording

process. With 78% of the users giving a rating of 4

or higher this was the least successful aspect of this

section. Despite some initial confusion by the users

they were able to understand how all worked and did

not find it to complex. In a general way the audio

response was positive.

Interaction

The main purpose of these questions about

interaction was to know users’ opinion about playing

sounds with their hands and the activities assigned

with each hand.

About the 3D position of the user’s hands,

specifically the number of action zones and

precision, it was asked whether 4 zones is the ideal

number of zones or if another number is more

appropriate. 61% of users said that 4 zones were

ideal, 13% chose 3 and 26% said that 5 or even 6

zones was preferable.

84% of users gave rating 4 or higher when asked

if it was easy to identify the zone in which their

hands were. There was an answer with rating 1 but it

was mainly due to the poor hardware of the

computer in which the test was performed.

About the choice of colours of the plans (green

and grey), 79% of users gave a rating 4 or 5.

Concerning the colours of the virtual hands (red and

blue) 86% of users gave rating 5.

Concerning the use of the right hand to control

the volume of the sound, as it does presently, or if it

should control another parameter, for example, a

sound effect like a distortion or something similar,

70% of users said that another parameter was

probably best.

Regarding switching the activities of each hand,

the answers to this question were analysed having in

mind the dominant hand of the user. Only 3 left-

handed participants were found and any findings

based on this universe of people are fragile.

Nonetheless, when all users are considered, 22% of

them would rather switch hand activities. However

if only right-handed people are considered this

percentage increases to 25%. Left-handed people

unanimously consider that the ideal is not switching

hands’ activities. This means that 34% of the

participants prefer to use the dominant hand to

compose music. However, it is worth mentioning

that most users said that it did not matter which hand

was assigned for each type of activity.

Visualization

The goal of the set of questions about visualization

was to know if the visualizations that were chosen

are adequate and if they were easily linked to the

corresponding sounds.

In the question about the clear matching of sound

with visualization, 70% of users gave a rating of 4

(easy) while the remaining 30% gave the maximum

rating of 5 (very easy).

Next it was asked how easy it was to identify the

visualization of the base track. 65% of users gave a

rating of 5. There were also ratings 1 and 3, but in

the case of the rating 1, the study was conducted in a

computer with poor performance. Some users did

not realise that the pulsating light at the end of the

scene matched the base track beat sound.

Regardless of track visualizations, users were

asked if the scene as a whole was enjoyable from a

graphical standpoint. 65% gave a rating of 4 and

31% gave 5.

Regarding the use of parallelepipeds (blocks) for

the visualization of normal tracks, 43% gave a rating

of 3 and 48% gave 4. It was also asked if they had

any alternative in mind for the visualizations. Most

GRAPP2015-InternationalConferenceonComputerGraphicsTheoryandApplications

460

suggestions were other geometrical choices as cubes

or prisms.

Lastly, it was asked if each block should have an

individual colour or, on the other hand, if they

should remain as they were. 78% of users preferred

individual colours. Some participants stated that,

besides colour, the order in which blocks are

presented facilitate their identification, therefore the

choice of colour is not so important.

5 CONCLUSIONS AND FUTURE

WORK

We presented an interactive Web application to

compose music from rhythmic sounds in real-time.

It integrates 3 main components: audio, to play and

record several sounds; interaction, using the Leap

Motion device to command music composition; and

visualization of the music that is played.

A user study was conducted involving common

users as well as experts to assess usability and obtain

user feedback. The interest shown on the innovative

nature of the application suggests an increasing

interest about new forms of interaction by the users,

especially of musical nature. Many of the users

revealed a great enthusiasm in future developments

and suggested the implementation of new features.

The main innovation of this application is to

combine the components of audio, visualization and

interaction, providing a natural interaction to

compose music with a Leap Motion device.

As future work it would be interesting to

combine more sounds and not only rhythmic sounds.

For example, sounds with notes as a xylophone

(discrete sound) or as a violin (continuous sound).

Using sounds like these ones will bring new

challenges, both in musical composition and in

interaction issues.

To improve interaction it would be interesting to

explore the new features of the more recent Leap

Motion driver. One of the new features is to identify

if the user’s hand is closed. This gesture can

command stopping playing a sound, eliminating the

need to use an action zone for this purpose.

ACKNOWLEDGEMENTS

We thank the Portuguese Foundation for Science

and Technology (FCT) and the R&D unity LabMAg

for the financial support given to this work under the

strategic project Pest OE/EEI/UI0434/2014.

REFERENCES

Airspace, Leap Motion Store – Airspace

https://airspace.leapmotion.com/

BeatPetite http://beatpetite.com/.

Bernhof http://bernhoft.org/.

Biquad Filter, The Biquad Filter Node Interface

http://webaudio.github.io/web-audio-api/#the-

biquadfilternode-interface.

Color Particles, Music Colour Particles. jabtunes.

http://jabtunes.com/labs/arabesque/.

Cube Visualizer.

http://shoffing.com/pages/projects/cubevis/.

Dive, A dive in Music.

http://do.adive.in/music/.

Dub FX http://dubfx.net/.

Leap, Leap Motion https://www.leapmotion.com/

Ljósið, Ólafur Arnalds - Ljósið (Official Music Video)

http://vimeo.com/6284199.

Loop, Loop Waveform Visualizer. Air Tight

Interactive.http://www.airtightinteractive.com/2012/01

/loop-waveform-visualizer/

Garcia, C. M., 2011, Study on the Procedural Generation

of Visualization from Musical Input Using Generative

Art Techniques. Master Thesis Texas A&M

University.

Panner, Panner Node Interface.

http://webaudio.github.io/web-audio-api/#the-pannernode-

interface.

Theremin, Ssergejewitsch, Theremin Leo. Theremin

Signaling Apparatus. US1658953 A USA, 14

Feb1928. Grant.

Tom Thum, Beatbox brilliance: Tom Thum at

TEDxSydney

http://www.youtube.com/watch?v=GNZBSZD16cY&

feature=youtu.be&t=8m12s.

Tone Craft.

http://labs.dinahmoe.com/#tonecraft.

ToneMatrix, ToneMatrix. AudioTool.

http://tonematrix.audiotool.com/.

Web Audio API.

http://webaudio.github.io/web-audio-api.

ZoeKeating, Zoe Keating - 'Lost' [HD] Sound Quality,

ABC Radio National. Youtube. http://www.youtube.

com/watch?v=sG9H5E2JN3s.

VisualizationofMusicatArm'sLength

461