Ssynth: a Real Time Additive Synthesizer With Flexible Control V. Verfaille, J. Boissinot, Ph. Depalle, M. M. Wanderley Sound Processing and Control Laboratory, McGill University

We developed a real time additive synthesizer called Ssynth, with advanced and flexible control functionalities. This research is a further development of Escher, a system developed for studying gestural control in interpolation of digital musical instruments playing. By considering synthesis from the control viewpoint, in terms of design and implementation, Ssynth allows to generate good quality sounds from an instrumental sound database, with a coherent control, providing interpolation and extrapolation of musical playing of digital instruments. For modularity purpose in the design of digital musical instruments, Ssynth is composed of two parts: a set of Pd patches that implements the different mapping strategies and layers, and a synthesizer1 implemented in C that can be compiled as a stand alone program or as a Pd object, using the Pd scheduler to have output audio. Two parameter conversion layers convert gesture data into higher-level parameters (fundamental frequency, intensity, and dynamics) and higher-level parameters into synthesis parameters. That way, the great number of additive synthesis control parameters is reduced to a smaller set. Additive frames are organized as a 3-dimensional mesh according to pitch (7 values), dynamics (3 values; related to loudness and brightness) and instrument (4 up to now). Morphing between N tones according to those parameters is provided in two steps: interpolated additive frames are weighted from pitch-shifted additive frames with different fundamental frequency and dynamics of the same instrument; then a morpher weights interpolated frames of several instruments. Morphing attack and release requires a time-warping of additive data to provide better quality timbre of the morphed sounds. The spectral envelope is a function of frequency2 that simplifies the control of partials amplitudes in Ssynth and is useful to morph sounds. When gesturally controlled, it may be necessary to convert a spectral envelope model into another, more suited to provide a stable spectral envelope for a given control. Ssynth uses various models and conversion methods between the following spectral envelope models: formants, cepstrum, LPC class (autoregressive filter, correlation function and reflection coefficients). The sound database contains additive analysis and spectral envelope models of wind, wood and brass instrument tones (clarinet and oboe as in Escher, plus saxophone and trumpet) from the McGill master samples database. Based on all those components and techniques, Ssynth allows for interpolating and extrapolating the database, synthesizing polyphonic sound, and handling OSC control messages.

References McAulay, R. J. and T. F. Quatieri (1986). Speech Analysis/Synthesis Based on a Sinusoidal Representation. IEEE Trans. on Acoustics, Speech, and Sig. Proc. 34(4), 744–54. Schwarz, D. and X. Rodet (1999). Spectral envelope estimation and representation for sound analysis-synthesis. In Proc. Int. Comp. Music Conf., Beijing, pp. 351–4.

1 It implements both 1-order or 3-order phase polynomial models (McAulay and Quatieri 1986) with scalar, vectorized and recursive formulation implementations. 2 Properties such as envelope fit and smoothness are added and tuned differently depending on the application (Schwarz and Rodet 1999).