`
`Jamie Ansticel, Tim Bell’, Andy Cockburnl, Martin Setchel12
`Department of Computer Science
`2School of Music
`University of Canterbury
`Christchurch, New Zealand
`j amie, tim, andy @cosc.canterbury. ac.nz
`msetchell @music.canterbury.ac.nz
`
`Abstract
`
`Computerising the task of music editing can avoid a con-
`siderable amount of tedious work for musicians, particu-
`larly for tasks such as key transposition, part extraction, and
`layout. Howevel; the task of getting the music onto the com-
`puter can still be time consuming and is usually done with
`the help of bulky equipment. This paper reports on the de-
`sign of a pen-based input system that uses easily-learned
`gestures to facilitate fast input, particularly if the system
`must be portable. The design is based on observations of
`musicians writing music by hand, and an analysis of the
`symbols in samples of music. A preliminary evaluation of
`the system is presented, and the speed is compared with
`the altematives of handwriting, synthesiser keyboard input,
`and optical music recognition. Evaluations suggest that the
`gesture-based system could be approximately three times as
`fast as other methods of music data entry reported in the lit-
`erature.
`
`1. Introduction
`
`For many writers, a word-processor is inextricably linked
`with the writing process. The benefits offered by word-
`processors are numerous, and include the ease of editing
`(both content and representation), the ability to store and
`retrieve documents, and the support for sharing documents
`with others. Many writers would find it hard to carry out
`their work without computer support.
`Computers offer benefits to written music that are analo-
`gous to those of word-processors. Computers can also pro-
`cess music in a variety of ways that would be tedious if
`carried out manually, particularly tasks like key transposi-
`tion, part extraction, and layout. Several computer systems
`support musical input, but the process is tedious and time-
`
`consuming [ 121.
`In this paper we describe our work on Presto, a computer
`system for pen-based musical input. Our aim is to develop
`fast and efficient mechanisms for musical input that require
`only minutes to learn. The system is intended for com-
`posers, arrangers, music editors, and performers. It is partic-
`ularly valuable where portability is an issue. Currently the
`preferred means of music input is to use a synthesiser key-
`board in combination with a computer keyboard and mouse,
`which could hardly be considered portable. In comparison,
`the traditional manuscript paper and pen can be used any-
`where, including in rehearsals and while on tour. A direct
`application of a pen-based music input system is for the
`“Muse” digital music stand [5], a design proposal for an aid
`for musicians. Muse has an LCD display for music, along
`with a variety of aids for performance and rehearsal, such
`as automatic page turning, annotation using a stylus, and co-
`ordination with other musicians’ music displays. Pen-based
`editing of music would be a natural addition to the system.
`Our work is presented as follows. Section 2 reviews
`current mechanisms for musical input, and describes recent
`work on pen-based and gestural user interfaces. To ensure
`that Presto’s input language is both fast and learnable, Sec-
`tion 3.1 reports on the relative frequency of musical symbols
`across a variety of musical scores, and Section 3.2 analyses
`the music writing techniques used by musicians. Section 4
`derives a design proposal for Presto from the analysis, and
`describes a story-board evaluation of the proposal.
`
`2. Background
`
`Hardware devices for human input and output have tradi-
`tionally been limited to a screen for output, and a keyboard
`and mouse for input. This limitation is rapidly diminish-
`ing, and it is now common for entry-level computers to be
`able to support many input and output devices, including mi-
`
`0-8186-7525-Xl96 $05.00 0 1996 IEEE
`
`260
`
`BlackBerry Exhibit 1010, pg. 1
`
`
`
`crophones, loud-speakers, pen-based tablets, musical key-
`boards, and so on.
`In this Section we review current mechanisms for com-
`puter input of music, and we examine novel pen-based input
`mechanisms that we believe have the potential to improve
`human-input of music.
`
`2.1. Computer Input of Music
`
`When compared to English text, written music is a highly
`expressive notation. Although the number of symbols in
`text and music are roughly equivalent, written music is a
`two-dimensional medium while text is linear - music can
`express several concurrent streams in one physical area. En-
`glish text consists of the alpha-numeric characters and punc-
`tuation, and words follow each other in a linear stream. Mu-
`sic consists of an arlphabet of symbols such as crotchets
`(quarter notes), quavers (eighth notes), and sharps, but the
`semantics of the symlbols depend on their horizontalundver-
`tical locations. Much of the difficulty of musical input to
`computers is derived from music’s two-dimensional seman-
`tics.
`Several computer systems support musical input. Once
`the music is held in a computer format it can be processed
`in a variety of ways. Normally the resultant output of these
`systems is a graphical representation of the musical score
`and audio playback of the music. Current techniques for
`musical input are reviewed below: A thorough review of
`these techniques is presented by Anstice [ 11.
`Direct Keyboard Entry is achieved through ASCII music
`representation languages such as DARMS [6] and MusicTex
`[ 171. Users must leau-n aprecise syntax, and any limitation in
`the language’s gramlmar constrains its flexibility. The cogni-
`tive mapping from ASCII notation to musical score is weak.
`Direct Manipulatilon Music Input. The graphical user in-
`terfaces provided by most current music systems, such as
`Lime [ 111 and Finale [4] combine a keyboard and pointing
`device with a bit-mapped display to allow direct manipula-
`tion of musical symlbols. Users create musical objects using
`key-selections and menus, and the properties of the objects
`can be adjusted through direct manipulation. Editing musi-
`cal scores is well supported by these systems (the users di-
`rectly manipulate the objects of interest), but initial input is
`frustrated by the abstracted mapping between musical ob-
`jects and keystrokes.
`Musical Keyboard and Direct Manipulation. Most mod-
`ern music systems allow users to enter music by playing
`a music keyboard (connected to the computer. This allows
`fluid entry of lines of music. There are two major problems
`with this style of musical input. First, if the music contains
`multiple lines of music, either the system must disambiguate
`between the lines, or (more commonly) the user must play
`
`each line independently. Second, recognising the musical
`rhythm is difficult, particularly when multiple lines of mu-
`sic must be correlated. Recognition errors must be corrected
`using the direct manipulation editing facilities.
`Optical Music Recognition aims to minimise human in-
`volvement in music input. The musical score is scanned to
`a bitmap image, and the computer system attempts to parse
`the bitmap. Error rates in current systems are relatively high,
`and although initial input of the musical score is fast, the
`subsequent human editing of the score is slow and tedious
`[12]. This technique is also dependent on a printed ver-
`sion of the musical score being available. Bainbridge and
`Carter [3] provide a comprehensive review of Optical Mu-
`sic Recognition systems.
`Soundtrack Analysis is a current area of research which
`involves computerised transcription from recorded music.
`Monophonic translation has been achieved [ 141, but there
`are substantial difficulties in polyphonic translation [ 131.
`
`2.2. Pen-Based Input
`
`User demands for portability have resulted in small com-
`puters that have little space for a keyboard. Small keyboards
`are cumbersome and slow to use for text entry [18], so alter-
`natives are an active area of research. Pen-based systems for
`data-entry are rapidly developing, driven by their popularity
`with users [8]. They consist of a tablet, which is normally
`an LCD screen (of any size), and an electronic pen or stylus
`which is used to write on the tablet.
`Pen-based input mechanisms range from those that make
`no attempt to recognise the users’ free-hand marks to
`those that constrain the user to tapping characters on soft-
`keyboards [ 151. Wang’s Freestyle system [7], for instance,
`allows users to annotate documents in free-hand using an
`electronic pen. Most pen-based systems, however, use some
`form of recognition to convert the user’s marks and ges-
`tures into computer text which is more legible than hand-
`written text and consumes less computer memory. The style
`of marks that the systems can recognise is the primary dis-
`tinction between pen-based systems.
`Cursive hand-writing recognition is the most ambitious
`style of recognition. There is great variation between in-
`dividuals’ hand-writing, so systems that attempt to recog-
`nise cursive hand-writing must be trained to recognise each
`user’s style. Systems such as the Apple Newton [2] can be
`trained to achieve over 90% accuracy, and although the er-
`ror rate remains frustrating for extensive text input, users
`are surprisingly enthusiastic about relatively poor perfor-
`mance [8].
`Boxed character recognition. To ease computer recogni-
`tion, some systems constrain the users to printing discrete
`
`261
`
`BlackBerry Exhibit 1010, pg. 2
`
`
`
`characters within specific regions on the screen. These sys-
`tems display a small box (or series of boxes) into which
`users print characters. The character entry boxes slide along
`the screen as the user enters characters. This technique
`vastly reduces the potential variability at each recognition
`step (from the set of words to the set of characters), but
`there is still substantial similarity between certain charac-
`ters, such as “U” and “v”. Some systems require users to re-
`enter characters that are incorrectly recognised (with the risk
`that subsequent recognition may also be incorrect), while
`others support a “Tap-correction” in which the most likely
`set of matches are displayed, and the user selects the desired
`character [9].
`UniStrokes. Recognition errors reduce writing speed and
`disrupt writing flow. To ease these problems, Unistrokes
`systems [lo] use an abstract alphabet in which all characters
`are denoted by a single pen stroke: each character is recog-
`nised when the pen leaves the tablet surface. The Unistrokes
`abstract alphabet is designed to reduce the similarity be-
`tween the character symbols, and yet to maintain a partial
`mental mapping between the symbols and the characters
`they represent. Unistrokes also allows “Heads up” text en-
`try, escaping the need for the user to look at the tablet surface
`while entering data. This is achieved by having all unistroke
`symbols entered into a small static region of the tablet. As
`soon as the pen leaves the tablet, the symbol that the user
`wrote is recognised, the recognised character is presented in
`an output region that shows the linear stream of characters,
`and the symbol mark is deleted allowing the next symbol to
`be written on the same region of the tablet.
`The primary disadvantage of unistrokes is that users must
`learn an abstract alphabet. This effort is minimised by the
`partial mapping from symbols to characters, and Goldberg
`and Richardson [ 101 report that the complete alphabet is nor-
`mally learnt in about 10 minutes. Unistrokes supports an ac-
`tive reference sheet for learning assistance.
`Marking Menus. Marking menus [ 161 are pie menus which
`pop-up under the pen-tip after a short delay (about half a
`second). Users then select the desired option by dragging
`the pen into the appropriate region of the pie menu. With
`experience, users learn the direction to particular menu op-
`tions, and then selections can be made by flicking the pen in
`the appropriate direction without the menu being displayed.
`To support the large number of options required for text-
`entry, T-cube [18] uses hierarchical marking menus, with
`keyboard shortcuts.
`
`2.3. The Potential of Pen-Based Music Input
`
`None of the music input methods reviewed in Section 2.1
`allows input at a speed that is comparable with hand-written
`music on paper. Hewlett and Selfridge-Field [ 121 describe
`
`a study of music input times for Haydn’s Symphony No. 1.
`Using input through a combination of music and computer
`keyboards, the initial input took four hours and twenty min-
`utes with subsequent editing taking nine hours and twenty
`minutes, a total of thirteen hours and forty minutes. Using
`the Optical Music Recognition system SightReader the ini-
`tial input took thirteen minutes, but the subsequent editing
`took nine and three-quarter hours, a total of ten hours and
`seven minutes.
`We performed an experiment in whic
`copied a third of one movement of the s
`in Hewlett and Selfridge-Field’s study. Averaging over the
`three subjects, and extrapolating the time taken, the hand-
`written input took only 42% of the time that the Optical Mu-
`sic Recognition input method took. Thus handwritten input
`appears to be at least twice as fast as other methods of in-
`put. Currently the disadvantages (with pen and paper) are
`the lower quality and the lack of flexibility. A pen-based
`computer system would
`Our goal is to use pen
`prove the speed of musi
`
`3. Analysis of Music Symbols and Music
`Writing
`
`Goldberg and Richardson [ 101 identified three primary
`criteria for their pen-based text input system, Unistrokes:
`ease of learning, high distinction between the input symbols
`(to ease recognition), and fast writing speed. These design
`criteria are equally applicable to musical input as they are
`for textual input.
`Learning can be assisted by a natural mapping be
`the input symbol and the intended musical symbol. Writing
`speed can be assisted by alloc
`the most frequently occurring
`tion analyses the frequency of
`ural” mechanisms that comp
`The data collected is used
`pen-based musical input.
`
`3.1. Musical Symbols
`
`Farquhar, and the instrumentation ranged from a quintet to a
`full orchestral score. Table 1 provides an ordered summary
`of the 22 most frequently ccumng musical symbols
`seven scores.
`account for 5 1 % of
`Table 1 shows that no
`on the page, and are far more frequent that a
`vidual symbol. Of the 1242 quavers and shorter notes, only
`
`262
`
`BlackBerry Exhibit 1010, pg. 3
`
`
`
`Symbol Name
`J
`- Slurortie
`d
`.
`0
`
`Quaver & shorter
`
`Crotchet
`
`Minim
`Durational dot
`
`sharp
`
`Y >
`’ J
`- Minim res1
`
`Number
`
`%
`
`1242
`
`23%
`
`1069
`
`20%
`
`445
`392
`333
`249
`
`196
`188
`166
`148
`141
`133
`107
`103
`
`90
`74
`65
`54
`
`8%
`7%
`6%
`5%
`
`4%
`4%
`3%
`3%
`3%
`2%
`2%
`2%
`
`2%
`1%
`1%
`1%
`
`49
`6
`5
`
`1%
`0.1%
`0.1%
`
`4
`
`0.1%
`
`Piece 3
`65
`7
`53
`16
`13
`7
`4
`5
`2
`
`Piece 1
`105
`
`Piece2
`68
`
`105
`54
`15
`6
`3
`1
`
`Object
`Black Heads
`Holiow Heads
`Stems
`68
`Beams
`36
`Accidentals
`11
`Bar Lines
`5
`Clef Signs
`2
`Dots
`11
`Quaver Tail
`1
`Slurs
`15
`Crescendo
`1
`Textual Object
`11
`184
`229
`289
`Total Objects
`75%
`82%
`91%
`% Head. Beam. Stem
`241 sec
`Average Time
`334sec
`352sec
`5Msec
`396sec
`Max Time
`483sec
`Min Time
`264sec
`221 sec
`127sec
`1.9
`1.46
`0.83
`Average Time per Object
`Table 2. Timing for copying music in task 1.
`
`2. Harmonising - subjects were given a melody line
`(from Bach’s chorale “Freu’ dich sehr, 0 meine
`Seele”) and were asked to write a simple four-part
`harmonisation. This task is cognitively demanding,
`and only experienced musicians could be expected to
`complete it.
`3. Writing from memory - subjects were asked to write
`a short melody from memory (“God save the Queen”
`was suggested).
`
`A complete analysis of these tasks is presented in [ 11. We
`the observations in two categories, first the time
`taken to input the data, and second the techniques used to
`write the musical symbols.
`Timing. There was a clear correlation between the com-
`plexity-of the task and the time to complete it. Even in the
`copying task, which could be carried out mechanically with-
`out cognitive processing, the average time to enter each mu-
`sical object increased with complexity. The number of mu-
`sical objects, therefore, has a lesser effect on data-entry time
`than the density of the objects: Table 2 shows that in the
`copying tasks the average time per object for the first piece
`was approximately half that in the third. It also shows that
`overall the first piece was input much more quickly than the
`third, despite containing many more objects.
`Most of the time in task 2 was spent in thought. Only 24
`notes had to be added to the music to create the harmony, but
`the average time for the task was 315 seconds, producing an
`average time per note of 13 seconds.
`Writing Styles. There was no noticeable variation between
`the mechanisms that each of the subjects used for the three
`tasks: their music writing mechanisms were the same for
`
`4
`b
`f
`Text
`I
`
`t-3
`
`z
`n
`
`Sindebearn
`Natural
`Flat
`Crotchet rest
`Textual symbol
`Bar rest
`Quaver rest
`Tail
`
`Double beam
`Accent
`Fermata
`Semibreve
`
`Mixedbeam
`
`7
`-
`Semiquaver rest
`E Multi-barrest
`Table 1. Freqiuency of musical objects in a
`sample of printed instrumental music.
`
`103 had tails. The others were marked by beams, with an av-
`erage of 3.7 notes per beam. Annotations to notes, such as
`dots, slurs and ties, accidentals and beams, are the next most
`frequent, accounting for 29% of the symbols, Rests account
`for less than 6% of the symbols.
`
`3.2. Music Writing
`
`‘Natural’ music writing methods were analysed using
`video protocol analysis. Ten of the subjects were postgradu-
`ate and senior undergraduate music students, and one a Uni-
`versity music lecturer. All subjects carried out three music
`writing tasks which were selected to illuminate any differ-
`ences between music writing techniques caused by cogni-
`tive aspects of the itask.
`1. Copying - subjects copied three pieces of music dif-
`fering in degree of complexity, from printed musi-
`cal scores. Although it is possible to copy the mu-
`sic with minimal cognitive processing (people with no
`musical experience could be expected to complete the
`tasks), we were interested to see how the complexity
`of the music affected the musical input times.
`
`263
`
`BlackBerry Exhibit 1010, pg. 4
`
`
`
`copying, harmonising, and “composing.” There were, how-
`ever, major difference between the individuals’ music writ-
`ing styles. Variations in the writing styles of notes, stems,
`and beams (the most frequent musical objects) are discussed
`below-see Anstice [ 11 for a full review.
`All the subjects started writing note groups (such as qua-
`vers and semi-quavers) by drawing the note-heads first.
`There was variation in the techniques used to add stems and
`beams. Seven of the subjects added the stems to all of the
`notes (left to right), and added the beam (if necessary) as a
`final step. Figure 1 illustrates this process. Three of the sub-
`jects, however, added the stems to the left-most and right-
`most notes first, then added the beam, and then added the
`enclosed stems, as illustrated in Figure 2. One subject drew
`the stems on the left-most note, extending it into a beam, and
`finishing the single pen-stroke as the stem of the right-most
`note. Stems of the enclosed notes were added as a final step.
`This process is illustrated in Figure 3.
`
`Figure 1. Common method for drawing quaver
`groups.
`
`Figure 2. Less common method.
`
`__
`- t - & p 7
`ar
`
`1
`
`I
`
`1
`
`
`
`Figure 3. Rare method.
`
`4. Presto: Design Proposal
`
`Our studies of printed music and how people write mu-
`sic have been used to design a pen-based music data entry
`system. The system is called Presto, because it is primarily
`intended to facilitate rapid input.
`The design of Presto has involved a trade off between
`learnability, recognition accuracy, and speed. Learnability
`
`is affected by the simplicity of the mapping from gestures
`to musical objects, and the number of gestures that must be
`learned. The accuracy of recognition depends on the dis-
`tance between symbols, and how simple they are. The speed
`of input also depends on how simple the objects are, and
`will be best if the shortest gestures are used for the most fre-
`quent symbols. Since the primary motivation for a musician
`to move from pen and paper to a computer is usually to save
`time and avoid repetitious work, we have opted for a trade-
`off that favours speed. The learnability and recognition ac-
`curacy have not suffered unduly because the bulk of musi-
`cal notation is covered by a relatively small vocabulary (Ta-
`ble l), and so there are few symbols to learn, and they can
`be quite different to each other to give accurate recognition.
`In our initial design for Presto, only the more common
`symbols have gestures. In Section 3.1 we observed that
`notes, dots, accidentals, slurs and ties, beams and rests ac-
`count for about 86% of the symbols on a page, and so ges-
`tures are provided for this subset. The small vocabulary of
`common symbols is backed up by making everything avail-
`able through marking menus, so even rare symbols will have
`a gesture, which is simply the menu selection gesture. This
`is not mnemonic, but can be learned, particularly if one sym-
`bol that is normally rare is common in the piece of music
`being edited.
`Our study of musicians’ writing techniques have indi-
`cated that it is desirable to allow partially complete notation
`to be entered even if it is musically incorrect. For example,
`in the process of drawing the beamed notes shown in Fig-
`ure 1, noteheads are drawn without stems, and in the second
`stage there are four crotchets, which could cause the number
`of beats in a bar to be exceeded.
`We have chosen to use shorthand gestures for Presto be-
`cause there is considerable redundancy in conventional mu-
`sic writing. For example, a crotchet note requires some time
`to colour in the head, and add a tail of the correct height.
`These repetitious tasks are best done by the computer, so a
`simpler gesture is used which is instantly converted under
`the pen to the correct symbol. The gestures have been de-
`signed to be mnemonic to make learning as simple as pos-
`sible, although this is secondary to the goal of making input
`efficient.
`The gestures to be implemented in the prototype of Presto
`are shown in Table 3. This shortest gesture, a dot, corre-
`sponds to the most common object, a filled notehead with
`a stem. The pitch of a note is indicated by where the ges-
`ture is made on the stave. The direction of a stem can either
`be chosen automatically by the system, or alternative rules
`(such as always up) can be specified by the user in advance.
`The stems can also be changed afterwards. Different note
`durations are constructed mainly by modifying a crotchet.
`For example, a horizontal line through a stem halves the du-
`ration of a note (users should think of it as a beam), while a
`
`264
`
`BlackBerry Exhibit 1010, pg. 5
`
`
`
`Musical Symbol Gesture Effect
`I
`J
`J
`d
`
`1
`
`Filled note
`
`Filled note with stem
`
`Draw a dot to get a filled note with auto-
`matic stem generation
`
`Draw a stem to place a filled note and give
`stem direction
`
`Start on the pitch of the note, draw right,
`then left
`
`Minim
`
`Doubles value
`
`Start drawing on note or rest
`
`Halves value
`
`Start drawing on note or rest
`
`Raise note
`
`Flick pen from note upwards
`
`Lower note
`
`Flick pen downwards from note
`
`Add dot
`
`Add tail
`
`Flick pen left from note or rest
`
`Draw line over one stem
`
`Add beam
`Add slur or tie
`
`Join stems to add beam
`Draw from first note to last note
`
`#
`
`b
`
`b
`0
`
`*
`
`Add barline
`
`Pop-up rest menu
`
`Delete Objects
`
`Rest
`
`Draw from top to bottom of staff
`
`Scrub up-down-up-down (down-up- down-
`up also works) over an object to delete it
`
`Table 3. Main gestures in the Presto system.
`
`265
`
`BlackBerry Exhibit 1010, pg. 6
`
`
`
`right-left gesture doubles the duration.
`The prototype system is primarily intended for evaluat-
`ing the effectiveness of the gestures for input, so we have
`not designed a very sophisticated system for editing the mu-
`sic. The main editing operation available is deletion, which
`is essential, particularly while users are learning the system.
`This is provided by a vertical “scrubbing” gesture. Other
`editing, such as moving, transposing pitch, changing stem
`direction, and adding ornaments, will be done via menus.
`The objects to be acted on by a menu selection will be se-
`lected by drawing a closed curve around them.
`The proposed gestures have been evaluated by a story-
`board simu1ation.l One of the authors wrote out 28 bars (six
`parts) of the Haydn Symphony that was used in the previous
`experiments. The first 13 bars took 84% of the time taken by
`the subjects copying by hand, and the last 15 bars took only
`77% as the writer became accustomed to the notation. We
`expect that the task will be a little easier in the implemen-
`tation where the writer will have (fast) visual feedback, and
`audio feedback is also likely to be valuable. Standard user
`interface mechanisms for increasing the efficiency of user
`input (such as copy and paste) will also be available in the
`prototype.
`These figures indicate that although the Presto notation
`is faster than conventional notation, musicians can write re-
`markably quickly despite the amount of redundancy in con-
`ventional notation. Extrapolating from these results, the
`Presto system appears to be about three times as fast as the
`Optical Music Recognition results reported by Hewlett and
`Selfridge-Field [ 121. Techniques for music data entry have
`improved since Hewlett and Selfridge-Field reported their
`results, but a three-fold improvement is needed to be faster
`than the Presto system. Also, Optical Music Recognition
`systems require high quality input, where as the Presto sys-
`tem can be used with low quality scores, and also for com-
`position and transcription, where the original does not yet
`exist.
`
`The Presto system has been designed with the goal of be-
`ing a fast pen based input system., The statistics gathered in
`Section 3.1 show that relatively few objects make up the ma-
`jority of symbols on a page, which has suggested a system
`with relatively few gestures, combined with menus to insert
`the less common symbols. Because there are few gestures
`they are easy to learn, and recognition is simple.
`A storyboard evaluation of the system has confirmed that
`the gestures are faster than conventional writing, although
`
`An attempt at a Wizard of Oz experiment failed because the gestures
`could be completed much faster than the simulated output could be put in
`place.
`
`only by a margin of about 16 to 23%, which is partly a re-
`flection of how quick writing music by han
`also explained by the overhead of the cog
`required for any music writing task. The c
`the Presto system may even be lower than c
`ing once users are experienced with it, because it removes
`the need to consider details like stem directions and the lo-
`cation of beams. The quality of the output from the digital
`system will be superior to handwriting at the speeds we per-
`formed our observations, and the output is a lot more flexi-
`ble.
`A preliminary evaluation of the Presto system indicates
`that it is as much as three times as fast as alternative meth-
`ods for entering musical data onto a computer. Since it is
`relatively simple to learn, its main drawback is likely to be
`that it requires pen-based input hardware.
`
`References
`
`J Anstice. Design of a pen-based music input system,
`1996. Masters Thesis in Preparation.
`Apple Computer Inc. Newton: Apple Messagepad
`Handbook. Apple Computer. 1 Infinite Loop, Cuper-
`tino, CA 95014-2084,1995.
`David Bainbridge and Nicholas Carter. Handbook on
`Optical Character Recognition and Docume
`Analysis, chapter 5. World Scientific Publish
`pany (in publication), 1995.
`
`Coda Music Technology, 610 Bury Drive, Eden
`Prairie, MN 55346-1718. Finale 3.5, 1996.
`Orya Dasna, Christopher Graefe, Justin Maguire, and
`Derek Wahlia. muse Digital Music Stand for Sym-
`phony Musicians. Interactions, 3(3), May - June 1996.
`Raymond F Erikson. “The Darms Project”: A status
`report. Computers and the Humanities, 9, 1975.
`
`an, D Cooper, and S Levine.
`E Francik, SE Ru
`Putting innovation work: Adoption strategies for
`multimedia communication systems. Communications
`of the ACM, 34(12):53-63,1991.
`C Frankish, R Hull, and P Morgan. Recognition ac-
`curacy and user acceptance of pen interfaces. In Pro-
`ceedings of CHI’95 Conference on Human Factors in
`Computing Systems Denver, May 7-11, pages
`509,1995.
`D Goldberg and A Goodisman. Stylus user inter-
`In Proceedings of the
`text.
`m on User Inte~ace Softwcsre
`and Technology., pages 127-135. ACM Press, 199
`
`266
`
`BlackBerry Exhibit 1010, pg. 7
`
`
`
`[lo] D Goldberg and C Richardson. Touch-typing with a
`stylus. In Proceedings of INTERCHI’93 Conference
`on Human Factors in Computing Systems Amsterdam,
`April 24-29, pages 80-87. Addison-Wesley, 1993.
`[ 113 Lippold Haken and Dorothea Blostein. The Tilia Mu-
`sic Representation Language: Extensibility, Abstrac-
`tion and Notation Contexts for the Lime Music Editor.
`Computer Music Joumal, 17(3), 1993.
`[12] WB Hewlett and E Selfridge-Field. How practical
`is optical music recognition as as input method? In
`WB Hewlett #and E Selfridge-Field, editors, Comput-
`ing and Musicology: An Intemational Directory of
`Applications, pages 159-166. Thomson-Shore, 1993-
`1994.
`
`[ 131 J. A. Moorer. On the Transcription of Musical Sound
`by Computer. Computer Music Journal, 1(4), 1977.
`
`1141 M. Piszczalski and B. A. Galler. Computer Analysis
`and Transcription of Performed Music. Computers and
`the Humanities, 13: 195-206,1979.
`[15] A Sears, D Revis, J Swatski, R Crittenden, and
`B Shneideman.
`Investigating touchscreen typing:
`The effect of keyboard size on typing speed. Behaviour
`and Informution Technology, 12(1): 17-22,1993.
`[16] MA Tapia anid G Kutenbach. Some design refine-
`ments and principles on the appearance and behaviour
`of marking menus. In Proceedings of ACM Confer-
`ence on User lntegace Software and Technology, Pitts-
`burgh, November 14-17., pages 189-195,1995.
`[ 171 Daniel Taupin. MusicTeX - Using TEX to write poly-
`phonic or instrumental music, June 1995.
`[18] D Venolia aind F Neilberg. T-cube: A fast, self-
`In Proceedings of
`disclosing pen-based alphabet.
`CHI’94 Coqference on Human Factors in Comput-
`ing Systems Boston, April 24-28, pages 265-270.
`Addison-Wesley, 1994.
`
`267
`
`BlackBerry Exhibit 1010, pg. 8