Other Synthesis Types
From DSPWiki
Contents |
Multiple Wavetable Synthesis
Multiple wavetable synthesis refers to two simple methods: wavetable crossfading and wavestacking, which are distinguished by dependence on the existence of multiple wavetables.
Wavetable Crossfading: Generates time varying timbres by crossfading between two or more wavetables over the course of an event. This type of synthesis is also called Compound Synthesis, Vector Synthesis, and Linear Arithmetic (L/A) Synthesis. The first commercial synthesizer to implement wavetable crossfading was the Sequential Circuits Prophet VS, introduced in 1985.
Wavestacking: Basically like additive synthesis, only it uses a more complicated waveform than a sine wave, such as a sampled sound. Each waveform on the stack has its own amplitude envelope, so sounds can fade in and out over the course of an event.
Wave Terrain Synthesis
Also called two-variable function synthesis, it extends the principle of wavetable lookup to the scanning of three-dimensional wave terrains (WT). A scan over the terrain is called an orbit, and can consist of any sequence of points on the wave terrain. If the orbit repeats itself (is periodic), so will the output signal. Varying the orbit over time generates time varying waveforms.
Granular Synthesis
Granular Synthesis builds up acoustic events from thousands of sonic grains, which last only briefly (1 to 100 ms - minimum perceivable time for duration, frequency, and amplitude discrimination). The first sound granulator was constructed by Dennis Gabor in the 1940s, and was a sprocketed optical recording system adapted from a film projector. He used this machine to experiment with time compression/expansion without pitch-shifting. A granular representation of sound is also implicit in the windowing technique used in the STFT.
Each sonic grains is shaped by an amplitude envelope, usually a gaussian, quasi-gaussian, three stage linear, pulse, or narrow impulse envelope. Several parameters can be varied on a grain-by-grain basis, such as duration, envelope, frequency, and location in the sound file for sampled grains.
Phase Distortion Synthesis
PD synthesis was a term created by the Casio corporation to describe a simple modulation technique used in its synthesizers. PD synthesis uses a sine wave table-lookup oscillator, and varies the rate of scanning over the cycle. The overall frequency is constant, according to the pitch of the note, but the waveform is no longer a sine wave. As the amount of speeding up and slowing down increases, the original sine wave turns into a quasi-triangle wave, and finally into a quasi-sawtooth waveform.
Waveshaping Synthesis
Jean-Claude Risset carried out the first experiments with waveshaping synthesis (also known as nonlinear distortion). As in FM synthesis, Waveshaping Synthesis gives us control on the time-varying bandwidth and spectrum of a tone in an efficient way. The fundamental concept is to pass a signal through a "distortion box". In the digital domain, the distortion box is a function that maps any input value to an output value. This function is called the shaping function or the transfer function.
Chebychev Shaping Functions:
By restricting the signal to an unvarying cosine wave, and using Chebychev functions, one can produce any desired combination of harmonics in a steady-state spectrum. One advantage of using Chebychev functions is that we can guarantee that the output of the waveshaper is bandlimited, and does not fold over the Nyquist frequency.
Amplitude Normalization:
One drawback of waveshaping synthesis is that the output amplitude varies, so if we want independence between timbre and output amplitude, some form of amplitude normalization must be applied. Three possible types of normalization are: loudness normalization, which is computationally intensive; Root Mean Square (RMS) Normalization; and Peak Normalization which is the most efficient.
