Analysis of the transformation at the beginning of Region II
This analysis was performed in collaboration with Jerome Kohl, who carried out the described turntable experiment and brought valuable ideas into the discussion.
From listening I always had the distinct impression that there is cross-fading, i.e. fading out of one event (silvery rattle), and fading in of another event (human crowd, initially transposed to a higher than normal speed), below a metallic, grinding and yet also 'rattling' sound oscillation, lasting just long enough to allow for this transition.
This impression of cross-fading is supported by two observations:
1. In its initial stages the slowing down in no way resembles the sound of the previous slowing down of the 'silvery rattle', at about 12" into track 12 of the CD – a few seconds after the marking 26'44" in the study score.14) Rather, it seems that here there is no real slowing down of the 'silvery rattle' at all, but that it is simply faded out below the above-mentioned metallic grinding, which appears to cover up the fade-out (here the start of slowing down sounds much noisier and more grinding than in track 12).
2. At about 45–47 seconds into track 13 of the CD two distinct sound events are clearly audible, the human crowd – still at an accelerated speed here – and the metallic, grinding sound (which disappears at about 47 or 48 seconds), strongly indicating that we are not just dealing with a simple linear process of slowing down the 'silvery rattle' until it becomes recognizable as a human crowd.
Even though the metallic grinding is not specified in the study score, the score does confirm a cross-fade event – albeit a few seconds earlier than the point where it becomes audible in the recording. Starting at 35.5 sec. into Region II (35.5 sec. into track 13 of the CD), and continuing to about 41", the score shows a second, overlapping wobbly line added to the first (the 'silvery rattle'), and this is labeled "Überblendung" ("cross-fade"). So something is in fact being faded in at this point, at that same pitch level. At about 41", this new line begins a downward glissando. There is not much audible change of sound during that cross-fade, yet the score clearly does suggest that the event that slows down to produce the downward glissando is different from the 'silvery rattle'.
(Later, after the initial exposure of the human crowd as such, at normal play-back speed, there are further cross-fades noted in the score.)
Does the 'silvery rattle' indeed consist of greatly sped-up sounds from a crowd? The answer seems to be yes, but with a twist. When the 'silvery rattle' is slowed down considerably, by turning an LP of the recording at a speed of about 5 or 6 rpm on a turntable by hand, these sounds do in fact sound like a crowd of voices. Unlike my original assumption, the shouts are no distinct events; rather, it is like a crowd at a football match (a "statistical" roaring), but reverberated. Most importantly, however, the sounds were processed by causing the playback speed to wobble rapidly (possibly produced by winding the tape past the playback head by hand, with a deliberately wavering speed); this wobble becomes a dominant feature of the sound. This fluctuation in speed naturally also causes a constant fluctuation in pitch and is therefore a kind of 'frequency modulation'. The 'rattle' appears to result from this speed fluctuation – and not from separate events of shouting.
Considering all this, what is a likely scenario of the real events in this famous passage?
The 'silvery rattle' appears to be a sped-up version of a human crowd, yet heavily processed, especially with the speed fluctuations that seem the actual cause of the 'rattle' when played back at high speed. When played back at normal speed, this heavily processed version would not reveal itself as a regular recording of a human crowd. Therefore, a cross-fade event had to take place, where the same recording of a human crowd, yet without the processing (or another recording of such a crowd) had to be introduced into the mix, barely audible at first. It would be this recording that performs the actual downward glissando by slowing down, until its content becomes recognizable as the human crowd it is.
The metallic grinding that seems to be laid over part of that downward glissando (audible, yet not mentioned in the score) might function to cover up the termination of the 'silvery rattle' before – or while – the second, 'linear' recording of the human crowd performs the actual slowing down.
Thus, when the composer states, in Jonathan Cott, p. 135:
"when this sound goes downward continuously and slowly enough, you discover at a certain moment that these are human beings shouting. All the shaking metal sounds we mentioned before were nothing but these voices – little boys shouting 'Hi, come here!' – speeded up enormously,"
this may be correct, but with a twist – a twist that is not mentioned in the statement.
Back to HYMNEN essay