"Tongues of Fire" makes use of every kind of signal processing technique available for compositions on the computer.

"Tongues of Fire" explores the grandeur and bathos of the human condition through the medium of the human voice. The slightly angry, disgruntled, comic mutterings with which the piece begins are the source of all the diverse sonorities in the piece: sounds which might suggest drums, water, metallic resonances, fireworks or entirely imaginary materials or events. Sound transformations of many sorts are a key aspect of the work.



The work falls into three major sections. The first section lasts for about 10 minutes. It begins with two statements of the sound material from which the whole work is developed and immediately begins to sonically develop this material. It ends with a gradually granulating voice slowly changing into a clock tick.



After a brief pause the opening phrase is recapitulated, leading to a rhythmic variation of the same material which forms the basis of the 2nd section of the work. This explores (among other things) humorous glissandi resonances and the shredding of dense vocal material into a water-like substrate (13m 20s to 14m 40s). Eventually, after some machine-like events, the vocal material becomes exceedingly dense, transforming into sweeping (filtered) noise bands, which form the transition section into the climax of the work (approx 18 mins.) Here, the rhythmic material becomes strongly pitched (being filtered through combinations of filter-banks) and leads into the "fireworks" transformations which begin the third section, the Coda of the work, where fragments of the preceding materials are recapitualated before the restatement of the "theme" of the very last moment of the piece.

"Tongues of Fire" attempts to demonstrate the great wealth of signal processing techniques available to the composer using a computer. Hence, almost every available process is employed at some point in the piece. I have newly invented many of the processes employed for this work. These include "Spectral Tracing", ­ a process in which a windowed Fourier Analysis of thesound is made using the Phase Vocoder; the sound is then reconstructed on a window-by-window basis using only the N loudest components, the process itself is time-variable. Waveset Distortion ­ the time domain signal is divided into elements at each even zero-crossing, and these elements are manipulated in a variety of ways: for instance, each is repeated N times to produce a strongly coloured form of time-stretching which generates pitched "beat streams" within the sound; or N elements are replaced by just one of their number, extended in length to span the entire set of N. The "fireworks" transformation is produced by this procedure, etc.


Another process that is used, although less radically, is "Sound Shredding", a technique related to Brassage. In this process a fixed length of sound is cut into arbitrary length segments which are reordered at random, and this process is then repeated many times. The progressive application of this process is heard in the voices to "water"-like transformation from 13m 20s to 14m 40s.