The vocal gliss fragments are short signals appearing at different times during the work. Initial and final points of the glissandi are found from the same set of tam-tam analyses used in sound installation fragments. These partial trackings1 go through a chain of filterings.
Only partials longer than 120 milliseconds are used. This will strongly reduce the amount of information.
100 random partials are selected.
A range of possible notes for the singers go into a loop.
For each singer, only partials less than 25 cents away from a possible note are passed through.
Only one random partial is necessary for each voice.
These procedures only select and approximate 5 tones, one for each singer, there are no additional distortions or transpositions. Initial and final chords are found from 2 different extracts from these tam-tam partial trackings to form a basis for transitions.
A glissando is created between the two tam-tam chords, curves are used as pitch shapes during the transitions.
Orchestral envelopes of attack and decay are added to the glissando texture, the overall duration is adjusted.
The result will have more characteristics of glissando textures than of the initial tam-tam spectrums.
The raw materials contained intermediate tones for a useful MIDI playback. Not all of this information was necessary for a live performance.
The final score was simplified to only show glissando dirations. The very high vocal register of Stine Motland gave a character to this fragment.
Example 68: CD Vocal gliss 4
1 Done through the pm2 library for Open Music: http://repmus.ircam.fr/openmusic/libraries