The Arts in Psychotherapy Computer aided music therapy evaluation: Testing the Music Therapy Logbook prototype 1 system

Research indicates that music therapists are likely to make use of computer software, designed to measure changes in the way a patient and therapist make use of music in music therapy sessions. A proof of concept study investigated whether music analysis algorithms (designed to retrieve information from commercial music recordings) can be adapted to meet the needs of music therapists. Computational music analysis techniques were applied to multi-track audio recordings of simulated sessions, then to recordings of individual music therapy sessions; these were recorded by a music therapist as part of her ongoing practice with patients with acquired brain injury. The music therapist wanted to evaluate two hypotheses: one, whether changes in her tempo were affecting the tempo of a patient’s play on acoustic percussion instruments, and two, whether her musical interventions were helping the patient reduce habituated, rhythmic patterning. Automatic diagrams were generated that gave a quick overview of the instrumental activity contained within each session: when, and for how long each instrument was played. From these, computational analysis was applied to musical areas of specific interest. The results of the interdisciplinary team work, audio recording tests, computer analysis tests, and music therapy field tests are presented and discussed. © 2011 Elsevier Inc. All rights reserved. The purpose of the proposed system is to help therapists keep track of what happens in music therapy, taking into account changes in music; so that changes in music can be objectively monitored (and quantified) in relation to other types of information, such as descriptive, written notes. The long term aim is to provide therapists with an everyday practice evaluation tool, which can be used to write session notes, record music therapy sessions, and analyse music data (Streeter, 2010, pp. 195–196). This research was supported by a Music Therapy Charity Research Fellowship awarded to Elaine Streeter, and a White Rose Health Technology Innovation Proof of Concept award to Elaine Streeter and Andy Hunt of the University of York, Joshua D. Reiss of Queen Mary University of London, Richard Caley of the Mid-Yorkshire NHS Trust and Cath Roberts of the Sheffield Teaching Hospitals NHS Foundation Trust. ∗ Corresponding author at: Honorary Research Fellow in Music Therapy, Department of Music, University of York, Heslington, York YO10 5DD, United Kingdom. Tel.: +44 07786 155451. E-mail address: [email protected] (E. Streeter). 1 Present address: INESC Porto Instituto de Engenharia de Sistemas e Computadores do Porto, Portugal. 2 Present address: Sheffield Health and Social Services NHS Trust, United Kingdom. The study follows from previous collaborations between music therapists and engineers. Since the 1990s, music therapists have made good use of computers and computational analysis to assist them in their research (Streeter, 2007). The first music therapist to trial computational analysis of music data from music therapy sessions was UK therapist, Mary Abbotson. Abbotson co-developed the Computer Aided Music Therapy Analysis prototype with engineers Dr Andy Hunt, Dr Adrian Verity, Mark Hildred, and music therapist, Felicity North (Hunt, Kirk, Abbotson, & Abbotson, 2000; Verity, 2003). In Finland, Professor Jaako Erkilla has been developing and testing the Music Therapy Toolbox system – a research tool that analyses MIDI data produced when patients and therapists improvise on MIDI instruments (Erkkilä, 2007). In Israel, Dr Avi Gilboa has been developing The Map (Gilboa, 2007), a computer based notation system for mapping music therapy sessions. In addition, Benveniste, Jouvelot, Lecourt, and Michel (2009) integrated Wii technology within a prototype digital musical instrument along with French music therapist, Edith Lecourt. The Music Therapy Logbook approach differs from those above in that it is the first to apply computational music analysis to recordings of individual music therapy sessions in which acoustic percussion instruments were played simultaneously with a MIDI 0197-4556/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.aip.2011.11.004 Author's personal copy 2 E. Streeter et al. / The Arts in Psychotherapy 39 (2012) 1– 10 piano. The approach posed a set of technical challenges: unlike MIDI data, audio recorded from acoustic instruments is subject to interference from other sounds occurring in or around a music therapy room. However, to meet the needs of practicing music therapists it was decided to work towards delivering a system that can cope with such challenges (many music therapists prefer, or only have access to, acoustic instruments). The focus of this paper is therefore to present the first set of test results generated by the Music Therapy Logbook Proof of Concept Study. The researchers used existing recording equipment and applied existing algorithms; these were adapted to generate information relevant to music therapy evaluation. The paper discusses the results of tests designed to find, identify changes in, and keep count of musical events across a series of individual music therapy sessions. The recording method was tested by a music therapist working with patients in a neuro-rehabilitation unit. The analysis tests were performed by a computer engineer in conjunction with a music therapist. Further research and development is necessary before The Music Therapy Logbook system can be made available to music therapists, however, initial software interface development (for Music Therapy Logbook analysis software) has been undertaken; a description of the preliminary software interface design is published elsewhere (Streeter, 2010, pp. 194–2011).