In this section, we evaluate
the project, with respect to the goals identified in section 1.2.
They were (1) the creation of an accessible, computer-based visualisation
of musical instruments coupling interactivity through musical (MIDI)
input, and (2) the application of this software in the field of
of our implementation with the specifications of our design, as
discussed in the last three chapters, demonstrates the satisfaction
of the first objective. The application is stable in the VST environment
and the graphics models appropriately and realistically respond
to stimuli from MIDI input from the user and the sequencer. Through
the piano and flute extensions, we have demonstrated the suitability
and flexibility of the system architecture of chapter 3 and instrument
hierarchical model of chapter 4. The tutor system also functions
well, as specified in chapter 5, with clearly observable modes of
skill: beginner and expert.
attempts to engender platform independence can only be effectively
evaluated through the porting of MIVI to other platforms, which
is beyond the scope of this project.
Likewise, as is the case with most educational simulations (see
section 2.4.2), little or no quantative or statistical analysis
of the software is feasible  – response times to input are observably
adequate and graphics performance is a moot issue (see section 3.2.1).
Instead, we turn to the experts of the music field to give us feedback
from two different aspects – technical and educational. We note,
however, that no feedback has been provided by laypersons, to which
the application is intended. The author cites two reasons for this;
(1) the current unsuitability of MIVI – in its prototype form –
to the general public, and (2) the problems associated with locating
aspirant – or possibly even failed – flautists and pianists17
who have yet to identify themselves as such.
To obtain a sampling of expertise
from the music field, the author approached the University of York’s
Music Department, Music Technology Department and Music Society.
Response was enthusiastic and positive in all respects, with no
shortage of interested parties. Following specially-arranged demonstrations
of the MIVI application, their response was as follows:
- Prof. P. Main, Dept. of Physics, University of York
Lecturer and Researcher in the ‘Physics of Music’ and Flautist
- Dr. A. Hunt, Dept. of Music Technology, University of York
Lecturer in ‘Electronic Musical Instruments’ and ‘Multiple Media Techniques’
From a technical standpoint, researchers in the
field of music technology were impressed at the level of integration
between MIVI and the VST host environment – concurring that its
selection as platform was a well-informed choice. The successful
importing of external dependencies such as OpenGL into the environment
also fascinated them. They also believed that porting to other OS’s
should be a relatively trouble-free affair.
Furthermore, the principal methods
and algorithms of the code (such as fingering, and instrument definitions)
met with their approval as well. They were surprised to see the
almost direct translation of finger maps (code ref. 45) from the
illustrations of Boehm’s book .
Both agreed that the extension to
a flute model, in addition to the piano, reflected well upon the
architecture. More specifically, Dr. Hunt conjectured that the system
presented an excellent platform to qualitatively study the failings
of the MIDI specification, whilst also implicitly identifying the
necessary extensions. In this capacity, he mentioned a desire to
see the MIVI system extended to encapsulate a full complement of
GM instruments, though he conceded that sound effects, such as helicopters
(voice #126) would only add to completeness, rather than practicality.
Again, both scholars agreed that the project posed
the interesting question of how to represent breath pressure and
lip shape – important aspects of wind instrument performance.
From a music technology point of view,
both were sceptical of the software’s application in an educational
role. Prof. Main, in his capacity as a flautist, stated that the
erudition of flute fingerings came more intuitively from knowledge
of hand configuration rather than finger placement. Although iterating,
with reason, that this was his own conviction, this opinion recurs
in the feedback of the next section. The concept, however, inspired
him to picture a physical flute interface – similar to a MIDI flute
– where the computer automatically depressed the keys, as with our
model. In this situation, the aspirant flautist could simply rest
their fingers upon the generic flute finger locations, and allow
the computer to move the keys – and, by transitivity, the hands
– into the appropriate configuration. It relies on the unchanging
position of fingers on the flute and, thus, it remains to be seen
if such a device is permissible for other woodwind instruments,
such as the oboe and clarinet.
As pianists, Dr. Hunt and the author
identified several problems with the piano tutoring system and solutions
that would address them. Principally, the system, in its current
state, discouraged the use of the score – Dr. Hunt felt that due
to the inflexible timing of the instructions, angst over the appearance
of the next note request (shaded green) led to discomfort, and an
inability to relax in the environment, requiring constant attention
on the graphical model. He stated that he felt ‘separated’ from
We agreed two solutions that would
solve this problem. Firstly, the playback timing could be controlled
through user response – a delay, pending note activation, before
playback continuation, or simply an inverse connection of tempo
to error count (the more frequent the errors, the lower the tempo,
the easier the recital). Secondly, the introduction of a visual
lookahead device, where future notes gradually ‘fade’ into focus
before being required, possibly, turning green to signify their
activation. This would give the learner time to prepare for shifts,
etc. and also allow them to choose more optimal fingerings.
On a similar line, Dr.
Hunt, feeling that the system was too note oriented said, "I
don’t feel like I’m learning music – more like it’s a hand-eye coordination
test." Instead of a note-by-note approach, he advocated the
introduction of phrase learning, where a bar or phrase was played
back by the host, and then echoed by the student.
- Edwina Smith, through the Dept. of Music, University
Flute teacher and pianist
- Susan Franks, through the Dept. of Music, University
of York / Leeds
Flute teacher (interview conducted over the phone)
- Oliver Hancock, Dept. of Music, University of
Student of Music, piano teacher and blues pianist
- Chris Bluemel, Dept. of Music, University of
Student of Music and pianist
- Jennie Wrigley, Dept. of Music, University of York
Student of Music and flautist
From the consumer point of view, all parties were
impressed with the concept of the application. Sadly, Ms. Franks’
schedule did not permit a private demonstration of the technology.
Others, however, were astounded at the quality, accuracy and detail
of the graphical models.
the interview, Ms. Smith used her own flute to compare the fingering
response of the simulation and estimated that, as a Boehm flute,
it was optimal and accurate. She did, however, mention that the
model was largely outdated and that a couple of extensions, namely
thumb lever and the Dorus key would be necessary additions before
successful application in the educational sphere was plausible.
The architecture of the model permits such amendments with the minimum
of fuss, simply through the alteration and addition of parametric
data. She also noted that, given the Briccialdi extension, the lever
would require a slightly higher degree of key-shape accuracy, for
the learner to be able to equate it to a standard modern flute.
On the subject of alternative and optimal fingerings,
she noted that many were only a requirement of advanced play, and
that most beginners are initially taught only a default fingering
for each note. She agreed that the ability to toggle this feature
would be of benefit to the beginning flautist.
Both flute teachers echoed Prof. Main’s
conviction that knowledge of hand configuration is preferable to
finger combination, and agreed that the addition of 3D hands and
fingers to the model would drastically aid the learning process.
During the demonstration, however, Ms. Smith conceded that our current
implementation still proved more attractive than the traditional
flute fingering tables of tuition books .
Oliver Hancock stated that hands would
be of use in the piano instrument as well. As a teacher of piano
with many younger students, he said that one of the problems he
faced was encouraging students away from the solitary use of their
forefinger. He also thought that placing the students in a room
with a MIVI system would show and guide them through this early
playing obstacle, and was of considerable advantage where students
are abundant and teachers at a premium.
In contrast to Dr. Hunt’s view (see
section 6.1.1), he said that the direction of the user’s gaze upon
the screen was actually of advantage, compared to the fixation on
the user’s keys and hands. Indeed, when play becomes competent,
the performers gaze should rest on the score, which is also separate
from the keyboard, and that the ‘separation’ from hands is a necessary
step to proficiency.
Also in contrast to Hunt, Hancock
appreciated the rigidity of the tempo, citing examples of young
students, when they use the score, incrementally stepping from note
to note without due consideration of their lengths. He feels the
feature would give learners an understanding of rhythm, though conceded
that this would be of less use to early learners. In the latter
case, he concurs with Hunt that the provision of phrase repetition
would be a good introduction.
As with the technologists, the musicians
would also be interested in an exploration of methods to display
lip shapes upon the embouchure and breath pressures. However, Ms.
Smith concedes that, although generic approaches do exist, many
people must find their own unique methods.