Publication details

Published online through the Centre for the History and Analysis of Recorded Music (CHARM)

ISBN 1-897791-21-6

London, 2009

© Daniel Leech-Wilkinson 2009

Version history

First version 1.0: 2009. Current version 1.1 (minor corrections): 14th February 2010.

How to cite this book

Daniel Leech-Wilkinson, The Changing Sound of Music: Approaches to Studying Recorded Musical Performance (London: CHARM, 2009), chapter no., paragraph no. [web address].


Daniel Leech-Wilkinson, The Changing Sound of Music: Approaches to Studying Recorded Musical Performance (London: CHARM, 2009), chapter 1.2.1, paragraph 22,


Note about online publication

Publishing this book online seems to require both an explanation and a certain amount of faith. There is still a strong feeling among academics that the excellence of a book can be immediately assessed by the name of its publisher, and that nothing that does not appear within hard covers can possibly be worth reading. This needs to change. Books (if one can still use that word) are for readers, and it is quite unreasonable to ask the reader of a book like this, who may well be a student or an underpaid musician, to invest (as buyers of my last two books were required to invest) £60 ($85/120) or more in order to have a copy on hand for future reference. Almost all this sum remains with the publisher and distributors. The return for an author, if he already has a university salary, is almost too small to register when spread over the years it took to write, especially as, given the exorbitant price, so few copies get sold.

In the past, one simply accepted this as inevitable. Few publishers were interested in musicology, and one felt lucky to be published at all. But the situation now is entirely changed by the internet. A book like this can be produced cheaply and conveniently without ever reaching paper, and its accompanying sound and data files can be downloaded immediately from the text. Readers can have it far more cheaply, or for a voluntary contribution, or for no payment at all. The advantages from the user's point of view seem overwhelming.

At any rate, this is my thinking in issuing this book online. I also want to see what happens. Will it be reviewed, cited, used, imitated, or ignored?

The book has been anonymously reviewed in the usual way, the reviewers’ comments leading to valuable improvements in several places. The XML tagging has been undertaken by Carol Chan, whose administration of the AHRC Research Centre for the History and Analysis of Recorded Music has been so impressive over the past five years; and the web implementation has been managed by King's College's Centre for Computing in the Humanities, where I am especially grateful to Harold Short, Paul Spence and Elena Pierazzo.

The text was completed in December 2008. I hope to be able to revise it from time to time. Until then, updates and corrections will be added when possible to the 'Updates' section. Comments and questions should be addressed direct to me: I very much hope that you enjoy the book.



If you read books about music I imagine it’s because you find music wonderful and want to know why. Although that fundamental question can easily get lost beneath the more immediate concerns of expert knowledge and practice, even if you are a music professional you got into music because it did things for you that nothing else can. So I’m hoping it’s safe to assume that most readers will agree that the most interesting and important question we might ask about music is ‘how does it move us’? Thanks to advances in musical science we recently started to get a lot closer to being able to answer that. There’s still a very long way to go, but at least we are beginning to get a much better idea of what goes on in our minds as we hear music.

Music doesn’t happen ‘out there’, and even to classical music we are not passive listeners. Music happens, in fact is made, individually, for each of us, in our minds as we listen, and that applies not just to listeners but to performers and (if they compose with sounds rather than shapes or formulae) to composers as well. Composers and performers have one or two other things to do at the same time—playing or singing, organising and writing things down—but it’s the listening that’s crucial, that for performers and composers determines what happens next and that for listeners determines how it’s understood. Because music is made in our minds, the place to look if we want to understand how it works is in the human brain. Mind is what the brain does, and because we’re now making new progress in studying how brains respond to sound, including musical sound, we’re beginning to get some really plausible answers to the most fundamental questions about what music is and what it does for us.

Nothing comes across more clearly from this work in musical science than that the performer is the source of all the most specific musical meaning. What the composer writes matters very much, but it’s what the performer does with that that shapes our responses, indeed that allows us to have responses at all. And so it’s time we looked much more closely at what performers do with scores. But how? The point of this book is to make some suggestions. I want to offer some ways of studying performances. For us to be able to examine them closely they need to be recorded performances. And therefore, so that we know what we’re dealing with, we have to spend a bit of time first of all looking at recording and where it can and cannot be trusted as evidence of performance. And before that we need to think a little about performance and its relation to compositions. So there are three preliminary chapters: one introducing the rather new field of the study of performances, one looking at the relationships between composition and performance and between performance and recording; and one outlining the way recording has changed so that we can better understand the worth of the sounds we hear as evidence for the practice of performers.

One of the things that’s immediately apparent from studying recorded performance, and that turns out to be absolutely central to understanding music-making, is that performance changes: in other words, there is such a thing as performance style. We can think of a performance style as a manner of making music that is temporary, only apparently stable, that changes gradually all the time, and that responds to changes in the wider world. It’s entirely thanks to recording, and to our having so many decades of it to hand, that we can now see for the first time, and think about, this gradual change in performance style. It’s one of the strangest things about music, for it suggests that much of what we assume is essential to musical performance could actually be quite different, was once, and will be in the future. Because the implications are so far-reaching we need to look at how style has changed really quite closely. And so the following three chapters are mainly about that. First (chapter 4) we look at twentieth-century singing, then at violin playing, and then at piano playing. Along the way, as we look at some performers and recorded performances in detail, I illustrate some of the more straightforward ways in which we can study the things performers do with musical sound in order to be expressive. At the end of this section (chapter 7) we take stock by looking at performance style from a more theoretical viewpoint, and try to understand why it changes and how. That takes us right to the heart of the question of how music moves us, because it requires us to think about what music is for, what it does to humans and how it relates to the way we experience the world. And that in turn puts us in a much stronger position to understand why performers change sounds from moment to moment, and why those changes move us. It’s that process that we examine in detail, using techniques involving computer visualisation of sound, in the last main chapter, chapter 8, which deals with expressivity and expressive gestures, first in singing and then in instrumental playing. The final chapter, chapter 9, looks briefly at where research might go next.

All the case studies in this book concern western art music, since that is where I work, but the approach is applicable far and wide and I look forward eagerly to seeing it extended. Most of the examples―by no means all, but most―are of singing, and most of the singing is of Schubert songs. I explain at the beginning of chapter 4 why singing deserves all this attention, so here let me explain why there’s so much Schubert. Schubert songs were recorded from the earliest days of recording, so we have over 100 years of examples representative of the musicianship of their times. There are a great many very different performances of the same pieces to compare. They are manageably short, and every one a joy to hear. They have texts, and so we can get some idea of what it is that their performers are intending to express―which is absolutely vital given that their way of being expressive is often unimaginably strange to us today. And Schubert song recordings from across the century are widely available in CD reissues and online, which means that you can easily find more examples to study. As we go along I try to identify further tests or exercises that readers might do with the recordings and data I provide, and I try as well to identify areas in urgent need of research. I hope the book will provide things for readers to do, if they wish.

Using the sound files

The sound examples I’ve provided are almost all drawn from original 78rpm discs in the archive at King’s College London. The advantage of this is that I can provide you with transfers knowing exactly how they were made. I cannot stress strongly enough how important this is, though it will become clearer from chapter 3. Recording introduces a host of falsifications (albeit mostly unintentional in the days of early recording), and transferring early recordings to digital files introduces many more. One cannot know, using a commercial transfer on CD, how far this has gone. So using original issues and transferring them according to simple, documented procedures is essential if one is going to be able to say anything about the performances that is not immediately open to doubt on account of the uncontrolled source. Except where a published reissue or other engineer is credited, all the transfers studied here have been made either by me or by my colleague Andrew Hallifax in the transfer studio at King’s College London. The settings we have used are documented as far as possible within the metadata for the CHARM online discography, where all these tracks can also be found. So unless stated otherwise, copyright in the transfers belongs to King’s College London. All other sound files are used either with permission or having extracted no more than 10% of the original track or item, in which case they are used under the provision in the Copyright Designs and Patent Act 1988, Section 30, allowing fair dealing for purposes of criticism or review. If you wish to reuse any sound files included here please be very careful that you are not contravening any rights.

Using Sonic Visualiser

To get the best out of chapter 8 it would be a good idea to download and install the software spectrum analyser Sonic Visualiser. Instructions on loading and using Sonic Visualiser can be found at The numbering of the SV files provided with this book indicates which of the sound files is required. If SV asks you to locate the .wav file, choose the sound file whose name begins with the second number in the SV filename. Thus the SV file requires the sound file dlw22.wav. As time goes by, however, SV will be outdated, and in that case you should of course use the best that there is and use it to study the sound tracks in a more sophisticated way than I can manage. I look forward to seeing the results. The best thing that could possibly happen to this book is that people adopt and then adapt its approach as fast and as far as possible so that it quickly becomes passé. It is intended as a starting-point, a spur to new research, and the sooner it’s superceded the more delighted I shall be.



During the several years I have worked on this book I have received funding from the Arts and Humanities Research Council, first through its Innovation Awards scheme and then through the AHRC Research Centre for the History and Analysis of Recorded Music, CHARM. It would be hard to list all the ways in which CHARM has contributed to my thinking about performance on record. Its four research projects have each had a substantial influence. Nicholas Cook and Craig Sapp, whose computing has extracted such interesting results from a vast amount of data on Chopin mazurka recordings, have shown how data collection and analysis can generate findings of real musical and music-historical significance. John Rink, assisted by Neta Spiro, were only beginning their project as I finished, but Rink’s underlying idea about performance motifs has been with me all along and surfaces here at several points. Eric Clarke, David Patmore, and Nick Morgan have helped me to understand more about music psychology, the record business, and collecting (a multi-talented team). Mark Sandler and his signal-processing research team at Queen Mary listened to my fantasies for an ideal sound analysis program and then wrote it (almost all of it) with apparent ease. Francis Knights and Andrew Hallifax have supplied me with vast quantities of discographical information and transfers from 78s which I could never have produced on my own, and Renee Timmers, with meticulous skill, taught me how musical questions could be tested experimentally. Watching over us all, advising modestly and wisely, has been Timothy Day whose knowledge and understanding of recorded music is without peer.

I owe a huge debt to Roger Beardsley, which he intends me to repay in a pub-crawl neither of us could survive were it proportionate. Roger has guided me over many years in everything to do with the transfer of 78rpm discs. Most of my text in chapter 3 dealing with disc manufacture and transfer was rewritten by him and then again by me, and so should be considered jointly authored, with him responsible for all that’s correct. He also provided quite a few of the recordings used here, including several I couldn’t find but that he was able to produce with a well-placed phone call. Denis Hall and Rex Lawson have shared their unique understanding of reproducing pianos, pianolas and rolls. George Brock-Nannestad, whose remarkable studies in the history of recording are cited in chapter 3, has provided a model for rigorous scholarship; as has Simon Trezise for investigative curiosity. Robert Philip got the whole field started with his 1992 Early Recordings and Musical Style, and it’s been a delight to discuss some of these issues with him. Karsten Lehl, whom I’ve never met, has given me over 500 transfers from his own collection and a list of all the Schubert song recordings (about the same number) in ours. He encapsulates the generosity that one finds everywhere among enthusiasts for recorded music and that makes this work such a constant delight.

Of the musicologists who have helped and inspired me I’d like especially to mention Daniel Barolsky, author of a fine PhD from Chicago on recorded piano playing; Rebecca Plack, whose pioneering Cornell dissertation offers a model study of early recorded singing; Nicholas Bannan, author of a remarkably bold doctoral thesis on music and evolution; and José Bowen, who needs no introduction to readers in this field since he was one of the first, after Robert Philip, to write about it. Dorottya Fabian has provided many helpful comments and suggestions. Colin Gough is a physicist, and one of the foremost authorities on the acoustics of violins and their implications for performance: he’s upset several of my findings, which is what scientists are supposed to do. I’m also proud to be able to thank my research students in early recorded performance, Abigail Dolan, Amy Blier-Carruthers, Miriam Quick, and YuanPu Chiao, more of whom will surely be heard shortly. And I owe much, too, to successive years of students on the King’s College London MMus course who have contributed to my thinking through ‘Schubert Song on Record’, and to undergraduates, especially those interested in music psychology, who took ‘Performance Practice on Record’, and ‘The Philosophy and Psychology of Music Perception’.

To King’s College London, and the then vice-principal Barry Ife in particular, I owe the extraordinarily bold decision to take on the BBC Gramophone Library’s duplicate 78rpm records. The archive has made a huge difference to my work and I’m enormously grateful. I was talked into talking King’s into it by Mary Gifford and in that sense (though it would astonish her to know this) she is probably more responsible than anyone for this book.

My greatest personal debt is to my children who have foregone more than one summer holiday while I got on with writing. Last time I dedicated a book to them it was mainly in middle French. I’m hoping this one may a bit be more useful.