Interesting discussion. The art and science division is not so hard and fast as implied here. There’s a blend of knowledge useful to the artist and scientist drawn from acoustics, psychoacoustics and perception at work that are all useful in guiding the actions and decisions of the composer/orchestrator. There’s always an interaction at work between the objective, measurable aspects of sound and their subjective, perceptual outcomes. For pretty much every aspect of what we do this division between measurement and perception exists and that’s where science and art blend — a useful blend to keep in mind, and pretty powerful too when you start to leverage that interaction in the making of well-crafted art, which is anything but "anything goes".
From my own experience in both audio engineering, music production and composition, I can add a further ‘rationalization’ of how Natural Volume works in relation to ensemble balance. When the sound of an instrument is captured in recording, the general goal is to record it at the highest level possible without distortion. This gives us the best quality sample with the highest dynamic range and lowest signal to noise ratio. But, as a consequence, quiet instruments, when played back at full level, will sound unnaturally loud in comparison to loud instruments, for which the recording level has to be lowered so as not to overload the recording system. This reveals the logic behind the ‘Natural Volume’ system. If you take a look at the table of Natural Volumes for the VSL instruments Dietz referred to, you can see that the quieter the instrument or ensemble is, the more reduction in its level is applied as ‘Natural Volume’. Instruments in the violin family are among the quietest in the orchestra, which is one of the reasons one need so many in a section to strike the desired balance in an orchestra in which almost every other instrument is significantly louder. To achieve a doubling of perceived loudness you need to increase the level by about 6 dB. Simply doubling the number of instruments won’t do the trick because adding t wo sounds of the same intensity only results in a 3 dB increase in level. So there’s a counterintuitive relation at work here between the number of instruments and their perceived loudness that also need to be taken into account. Art informs science.
The timpani is, indeed, an instrument with a huge dynamic range and can generate among the loudest sounds in the orchestra when played hard. Anything with that size of vibrating diaphram and corresponding resonator is gonna be loud! This is fundamental orchestration technique. Things get complicated in the age of electronic recording and mixing. The technical aspects of electronic/digital audio production can cause these real-world details to be lost. Before the age of audio production the two basic controls over balance were the interaction of the acoustics of the concert hall and the group of players under the direction of a conductor. Now, in the age of audio production, if you want the timpani to be present at lower levels a judicious application of EQ can help bring it out without resorting to playing it louder and overwhelming the ensemble of quieter instruments. This may not necessarily be ‘Natural’ according to the laws of acoustics that determine the sound an instrument in a particular hall, but may sound more ‘Natural’ to the ear. Where does one draw the line?
In addition, players will naturally modulate the intensity of their playing based on the acoustics of the room they play in. If you recall, all the VSL samples have been recorded in one and the same neutral acoustic environment. That’s not the acoustic environment you’re composing, arranging, and orchestrating for. We put our samples in one of several virtual spaces based on either the venues provided by MIR or some other spaces based on impulse response recordings, or even algorithmic simulations of resonant spaces. The loudness of each and every instrument is altered by the resonant characteristics of the particular space it’s being sounded in. They *will* sound different in relation to each other in different spaces depending on the relation between the frequency content of each instrument and the resonances specific to the space they’re sounded in. In this sense Natural Volume really is only a starting place, based on the ‘ideal’ acoustic of the silent stage, an acoustic space that’s been treated to minimize such resonances and their ability to alter the overall presence of an instrument.
Two cents worth of the art of science, or is that the science of art?
Best,
Kenneth.