Learn To Play explores the concept of ‘playability’ of music as a means to help those learning an instrument (flute, guitar or renaissance lute) by developing an online system to estimate the difficulty of a displayed piece, grading it according to accepted pedagogical standards.
Eventually the system will create on demand a set of helpful practice exercises based on player-selected passages within a piece or those that the system judges to be tricky.
The guitar is the most widespread instrument in the world today, and the internet provides a literally bewildering number of ‘tabs’ (scores notated in the format known as tablature) which do not require a formal knowledge of music notation. Tablature provides graphical instructions about the placement of fingers to form chords or melodies and the sequence in which they should be played. This is a system that has stood the test of time. Tablature has been used in parallel with standard music notation over hundreds of years, at least since the 15th century, and is particularly useful for instrumental teaching, especially in the early stages.
The Learn To Play project team from will investigate ways to help musicians find music to suit their level of attainment from the vast amount that is available online. This will be done by analysing the level of playability of tablature versions of pieces of music for guitar (classical and other styles) in machine-readable encodings and for renaissance lute (of which the research team already has a medium sized corpus of c10,000 encoded pieces collected in the ECOLM project). We shall use measures based on the hand-stretches and position-shifts indicated in the tablature to compute indexes of playability of both individual chords and transitions between adjacent pairs of chords. The resulting estimates of lute and guitar music playability will be assessed by both professional and amateur players as well as teachers.
One of the most popular instruments played in schools today is the flute, so we shall also build on earlier work carried out by co-I Fiebrink on the modelling of difficulty in graded pieces of flute music. This forms a very useful preliminary study, since the monophonic texture of the music allows us to focus on its time-based melodic aspects rather than having to account for chord-playing (as on guitar or lute). The revised model will be assessed by a panel of flute-teachers and players.
We shall then use standard machine-learning techniques to build models of playability which can be used to identify difficult passages within unknown pieces. These will be evaluated in human assessment by players of various levels of attainment and by guitar and lute teachers. The models will also be used to grade pieces (based on the difficulty of the most technically-challenging passages) and the results compared with the grades listed by music publishers in their catalogues.
Having found the most awkward passages in pieces, the proof-of-concept system we shall develop as a demonstrator forming the main output of the project will use simple algorithms based on discussions with teachers of the instruments concerned and with the extensive musical literature of pedagogical material to generate entirely new exercises derived from these passages for practising by a student.
To be most effective and acceptable to users, the music needs to be presented within a high-quality graphical user interface. Throughout the project we shall work alongside a music-industry partner, Tido Music.