This text is an abstract for a paper on a research in voice analysis as well as a proposal for the production of an interactive audiovisual installation on the theme of rhetorical musicality. The installation is driven by a sonofication of data from the study of a large set of recordings of political speeches.
The musical aspects of speech-making go beyond a Cicerean pronunciatio; indeed classicist composers skilfully employed rhetorical form, tropes and figures to write 'convincing' music. In the in the 20th century, the use of such means became associated with totalitarian regimes, and was consciously abandoned by progressive musicians. Likewise, contemporary politicians, at least in democratic fora, by default resort to a mode of delivery that Aristotle would have called iskhnos, "dry". What is the role of musicality in today's speech-making? Can we understand something about the power and attraction of contemporary orators through music analysis of their vocal delivery?
These questions are relevant to my research in composition as a metaphor for rhetoric which is expressed in pieces such as Khreia (2001), ReTreTorika (2005) and TreeTorika (2006) as well as in several published articles. After having worked with recordings of Mao, Hirohito and Palme, I am now preparing a large-corpus study of recordings from the General Debate at the opening of the General Assembly of the United Nations 62nd session, held in September-October 2007. Just short of two hundred 15-minute statements, given by presidents, prime ministers or other representatives of the member states, the dataset totals almost 50 hours of speech.
The recordings are analysed automatically using Praat (Boersma, Weenink 1993), MaxMSP (Zicarelli 1997) and OpenMusic (Agon, Assayag 1997). A number of musical parameters and their developement over the duration of each speech are measured: prosody, rhythmic obliqueness, phrase duration, voice ambitus, dynamic range, phrasal envelope, local variety of intonation, flow, shimmer, jitter and vowel differenciation. These linguistic-musical data are correlated with the age, sex and language of the speakers, as well as with statistics pertaining to the country they represent, i.e. size, population, GDP and age as independent nation.
The results will be presented in two forms. The research paper will focus on describing project situation, methodology, problems encountered and findings. The artistic presentation is an installation in a centre-axis symmetrical room with video screens and sound diffusion points distributed along the walls. The sonofication of the data is made in a semiautonomous real-time application influenced by continuous input from several camera monitors. The sound synthesis is achieved with an adapted form of CataRT (Schwartz 2006). As the visitors move about, speech transforms into song, fluttering voices from a large unruly choir, and back again to speech. The further the visitors, as a group, are from the screens, the more amorphous, i.e. non-syntactic and song-like is the sound produced. When a visitor approaches a screen, the sound gradually 'focusses', locally, on one politician's speech. The imagery consists of videos of the politicians; however, the installation is entirely driven by the audio logic, and thus images follow the sound.