Thursday, 5 November 2009
This is exceptionally interesting.
What is happening here is that a chap has written a patch and designed a machine to mimic human speech patterns. The composer starts by recording a sound, in this case a child bizarrely reciting a document from the European Parliament. This sound can be analysed spectrographically, with particular regard to the speech formants, which are the resonant frequencies of various parts of the mouth and throat that allow us to create and distinguish the different vowel sounds. Speech generally doesn't have a specific 'tone', and so the defining character of the sound will be these formants, of which two are usually enough to distinguish any particular vowel. What the composer has also done is created a 'map' of the potential resonances of a piano. Each of the piano's 88 sets of strings vibrate at numerous different frequencies, in an arithmetic sequence (x,2x,3x,4x,5x... etc), diminishing in amplitude as the vibration rate increases (it is the clash of the logarithmic sequence of apparent pitch (octaves at x,2x,4x,8x,16x,32x... etc) and the arithmetic sequence of vibrations that makes harmony possible in the first place). By mapping out the potential frequencies of the piano as a sort of 'palette' of sound, and then constructing a machine that is capable of highly sensitive fractional nuance of tone, the original sound can be retroactively mapped onto the piano and then reproduced.
Now; this stimulates a number of little ideas we like; one - This ability to hear the voice through only partial recreation is known as 'auditory pareidolia', a sort of sonic 'gestalt' whereby a mostly random pattern is interpreted as a recognisable voice. Familiar as 'EVP', this is H-logy in a nutshell; the uncannily possessed technology (i.e. the machine inhabited by spirit, the non-present in presence etc etc.), which has been discussed at much length in various places.
two- one of the standard tropes in criticism of musical performance is what is often called 'the cry'; especially with woodwind, brass and string players, the ability to mimic human voice patterns is considered to be one of the most important paths to excellence of performance. What the video above suggests to me is that in the technique of 'great' pianists, there might already be barely conscious variations in fingering pressure that shape the overall tone slightly towards the timbre of vowel sounds, which would partially explain the emotional power of certain performances.
ps- this double coding of sound is something that I've done a fair bit of research into...