pigrem

The Influence of Recording Technology on Expressive Timing in Popular Music

Abstract

This paper presents an analysis of the effect of recording technology on expressive timing in popular music. The initial hypothesis for the work is that the increased mechanization of the recording process is slowly reducing agency for expressive timing in recorded performances. 

The democratizing power of recording technology makes it quicker and easier than ever before to create music, however what are the effects of commonplace tools and practices such as multi tracking, audio and midi quantization, loop based composition, and a ready supply of homogenized sample CDs and presets on the expressive values of popular music.

13 Uk No.1 singles have been analyzed spanning a period of 65 years. Expressive timing data for each piece is presented and discussed, and the unlikely conclusion that ‘Ed Sherran saves music’ will be proposed.

Introduction

Technology is ubiquitous in today’s music industry, and its influence is sonically evident to even the untrained listener. Beyond discussions of the democratization of music making, or the precision, speed or accuracy technology fosters, is it adding more than it is taking away?

The power of expressive musical performance is well documented, with much research suggesting temporal and dynamic factors to be key expressive tools. Great performances of great works are compared and analysed to try and rationalise and attribute the source of such greatness.

Contrary to much of this research, recording technology and common studio practice has been heading in an increasingly less ‘human’ directions for the last 65 years. This project investigates changing levels of expression in popular music, and parallels observed changes to developments in recording technology. 

13 UK No.1 singles have been selected as a data set for analysis. To ensure an un-bias selection, No.1’s form the last week in March (the week this paper was written) every 5 years from 1957 to the present day have been chosen. As the 1950’s saw both the development of multi track recoding and the birth of the single chats the period is contextually relevant and accessible in terms source material.

The key area of investigation is expressive timing, as the area most affected by common audio and midi manipulation. Performance time will be analysed for all pieces of music in the corpus and standard deviations ((tempo BPM) compared as a function of time.

Developments in Recording Technology

Recording for many years maintained a ‘fly on the wall’ approach to documentation, and a single microphone, carefully placed would collect the sounds of performers around it. From the mid 1950’s technology pushed the recording arts forward in many ways. Les Paul developed multi-track recording, facilitating individual tracking and later overdubbing of recorded performances. As synthesiser technology developed, automatic sequencing of musical and percussion parts was pioneered, and devices such as the RCA MK I and later Buchla 100, EMS Synthi 100 and Fairlight CMI provided electro-mechanical, digital and computerised playback of sequenced parts.

In 1984 MIDI changed the world of music performance and recording once again, and the protocol became the backbone for recording, editing and sequencing of music parts in popular music. From this seminal protocol, a host of DAW based recording platforms have grown, including the commonly used Pro Tools, Logic Pro Cubase, and more recently Ableton Live.

The cost benefit afforded to home recording solutions by digitalisation throughout the 90’s and 00’s has seen a huge democratisation of music making, and with it a host of quick and easy alternative to not being a trained performer saturate the market. Loop based approaches to composition, facilitated by temporally homogenised and heavily processed audio files, and platform designed for purely rearranging a collage of pre-set material prevail. I can’t help but wonder if along the way somewhere we forgot what music really was…

Data Set

The source material analysed to create the dataset is listed in the table 1.1 below.

Method

Tempo Processing

To extract timing information from the corpus of selected works, Simon Dixon’s beetroot plugin for Sonic Visualizer was used. The beetroot plugin functioned well, however considerable manual correction was needed in areas with slower note onsets, or to accurately map rubato and ritardando sections. 

Annotation

The timing information was further annotated in sonic visualizer. The analysis took a beat level (crotchet) approach so as to capture expressive performance within each bar. Quaver resolution was tested, however factors such as swing and shuffle were observed to influence the results.

Analysis

Once notated, the timing information was exported from Sonic Visualser and imported into MatLab where it was plotted and the standard deviation for each track extracted. A linear regression has been calculated and is presented below in figure 1.14.

Data Analysis

Results

The resulting data was plotted as performance time over score in MatLab, and a standard deviation for each track produced. The result are displayed below in table 1.2.

The graph below displays these results as a function of time and a linear regression model suggested.

Discussion 

The results were surprisingly informative throughout and illustrate the initial hypothesis of the paper effectively. In terms of the noticeable effect of recording technology on the expressive timing in popular music, the data clearly evidences some trends that parallel well to key development in recording practice.

The first four pieces analysed (figures 1.1, 1.2, 1.3 & 1.4 - 1957/1962/1967/1972) show clear human variation in tempo and the presence of expressive timing throughout. It should be noted the precision of the Shadow performance (1962) is notable, and potentially evidence of multi track recording. 

Moving through the 70s we can start to see the influence of electronic sequencing and drum machines. Figure 1.5 & 1.6 (1977/1982) show a much more regular approach to expressive timing, which could be attributed to electronic/digital sequencing. Both works have a clearly synthesised backing track, so sonically concur with this assumption.

As we reach the 80s, the use of Midi and MTC is very evidence in the analysis of Mel and Kim’s “Respectable’ (figure 1.7 - 1987). The regularity of tempo fluctuations observed is clear evidence of computer-generated expressive timing. With the exception of Shakespeare’s Sisters ‘Stay’ (1992) the trend for mechanised lifeless timing continues throughout 1997/2002/2007 as illustrated in figures 1.9, 1.10, & 1.11. 

The final two pieces analysed (Gotye and Ed Sheeran) both seem to buck 30 year trend for lifeless expressive timing in popular music, although still demonstrate the same degree of computed regularity as seen throughout the previous decade.

Limitations

Although the conclusion draw from the study are a good fit to the hypothesis, there are several limitations to the research. 

Firstly, the sampling process reviewed a very limited number of examples form each time period, so no real conclusions can be drawn without further analysis of a wider corpus of material. 

Secondly, the analysis does show strong stylistic relevance as well as time based relevance, and potentially there have always been a balance of approaches represented in the charts, however potentially missed by the selection processes discussed above. 

Conclusion

So to conclude, does Ed Sheeran really save music? Well maybe not, however he does provide evidence that todays top popular writers and producers do consider the value of expressive timing in their work. 

More work and research really is needed to make any stronger claims, and a much larger dataset of historic recordings will need to be analysed.

Timeline

Bibliography

Bowen, J. 1996. Tempo, Duration, and Flexibility: Techniques in the analysis of performance

Chew, E & Callender C. 2013. Conceptual and Experiential Representations of Tempo: Effects on Expressive Performance Comparisons

Dixson, S. 2001. Automatic Extraction of Tempo and Beat From Expressive Performances. Journal of New Music Research, Vol 30, No. 1, pp39-58.

Huber, D & Runstein R. 2009. Modern Recording Techniques. Focal Press: USA

Repp, B. 1992. Diversity and commonality in music performance: An analysis of timing microstructure in Schumann’s ‘‘Träumerei’’

Repp, B. 1993. On Determining the Basic Tempo of an Expressive Music Performance. Haskins Laboratories Status Report on Speech Research SR-114, 159–166. New Haven

Rumsey, F. and McCormick, T. (2002). Sound and recording. Oxford: Focal.