Realtime Analysis of Dynamic Shaping Jörg Langner, Reinhard Kopiez ...

Eric Cheng

Description

Realtime Analysis of Dynamic Shaping Jörg Langner, Reinhard Kopiez ...

Informations

• Publié par: yrokaleb
• Nombre de lectures: 74
• Langue: Français

Extrait

Eric Cheng

3.1.06

ISE 599

Realtime Analysis of Dynamic Shaping

Jörg Langner, Reinhard Kopiez, Christian Stoffel, Martin Wilz

In the field of performance research, musical dynamics is often a neglected parameter.

Instead, research has mainly focused on musical timing as the parameter of choice. It is unclear

why this is the case, but the authors conjecture that it could be due to a lack of adequate research

methods for studying musical dynamics. In this paper, the authors hope to fill this void by

presenting a method for realtime analysis and display of dynamics.

First, the authors present a non-realtime method based on the assumption that the piece of

music is available as a digitized audio file. From this audio file, the loudness curve in sones is

calculated using a computer program developed by Bernhard Feiten and Markus Spitzer. Next,

the loudness curve is subjected to a smoothing procedure in which a given loudness point is

averaged (according to a smoothing curve) with surrounding points within a window of a given

length. The choice of window length determines the scale of the phenomena being examined.

Smaller window length exposes more local variations in dynamics, while larger window length

reveals more global characteristics. Finally, the gradients in every smoothing curve are

calculated at each point in time to ensure that strongly smoothed curves can obtain just as steep

gradients as more weakly smoothed out curves. The paper was a little vague on the exact

procedure they used to achieve this. The resulting data are displayed in a graph showing the

gradients in a “Dynagram,” where the x-axis is time and the y-axis represents varying window

size. Red coloring signifies crescendo and green coloring shows decrescendo.

In the realtime procedure, the audio signal is recorded through a microphone to a

computer. The procedure is essentially the same as the non-realtime version, but there is a slight

delay, since the smoothing out procedure necessarily requires information from the future.

Longer delays are required for larger window size while the delays for small window sizes are

inconsequential.

The procedures were tested on performances by a professional and non-professional

pianist of Erik Satie’s

Gymnopédie No. 1

. The most interesting result was that the professional

performance showed a more intense and well-defined loudness pattern, which corresponded well

with the formal structure of the piece. In the non-professional performance, on the other hand,

the dynamic shaping of the Dynagram was less pronounced and did not reflect the formal

structure of the piece as clearly. In addition, Dynagram analysis of a Bruckner Symphony

revealed dynamic shaping on a larger scale, showing a dynamic build-up spanning 20 minutes

from the start to the final climax of the piece.

The authors conclude that the Dynagrams reveal important attributes of dynamic shaping

on many different scales. In particular, the absence of presence of significant dynamic shaping is

clearly visualized. They also note the Dynagrams potential usefulness as a teaching tool for

musicians. The Dynagram could be displayed realtime as the pupil plays, allowing the teacher

provide instant analysis of dynamic shaping.

In conclusion, the authors present a novel method for visualization of musical dynamics.

The Dynagrams offer a clear representation of dynamic shaping, allowing for qualitative analysis

of musical dynamics. While such visualization techniques alone could never make up for a lack

of musicality, any additional tools are welcome. It is always helpful to have novel ways of

representing and communicating musical ideas, especially when you are attempting to achieve a

musical interpretation in a group setting.