Wavetable synthesis
Common synthesis technique used in sampling synths.
Synth.wavetable — Functionwavetable(table :: Vector{Float32}, amp :: Signal, phase :: Signal)
wavetable(s :: Sample, amp :: Signal, phase :: Signal)
wavetable(s :: Sample, amp :: Signal, freq :: Real)
wavetable(s :: Sample, amp :: Real, freq :: Real)A simple wavetable synth that samples the given table using the given phasor and scales the table by the given amplitude modulator. A copy of the given table is stored. If you already have a Sample, you can use its samples by passing it in directly. The sample's loop settings are not used in this case.
The wavetable is a simple array of samples, often in the same sampling rate as the output audio end point, but not necessarily so. The phase signal, which can be generated using phasor, has values in the range $[0,1]$ with $0$ mapping to the start of the wave table and $1$ to the end. To be precise, if the wave table has $N$ samples, then the phase values in the range $[0,1/N)$ will map to the first sample at index $1$ and the sample at index $k$ will be chosen by phase value in the range $[(k-1)/N, k/N)$.
The wavetable processor does 4-point interpolation. This means you can take a small sample table and stretch it out using the interpolation by using a slowly varying phasor. This is how the "pitch" of the sample gets changed during synthesis.
The amplitude modulation is just as with sinosc.
The samples wave table is deemed to be completed when the phasor signal or the amplitude modulator signal are completed.
It is common for the output of such a wavetable "voice" to be modulated using an envelope like adsr. You can therefore pass such an ADSR signal as the amp argument. If the phase signal is infinite in extent, then the lifetime of the wavetable voice will become determined by the lifetime of the ADRS envelope.
A complete voice implementation using a wavetable synth will need some extra niceties like a leading fragment of sound used during the "attack" phase of the voice until the voice begins to systain, and perhaps switching to a different voice during the release phase with a bit of cross fade. The current implementation provides a primitive that can be composed with other units in order to make such a more complete voice. The complexity of computing this voice will then be a bit more than the basic wavetable voice, but not much more than if you'd hand coded it yourself, thanks to Julia.
The phasor determines the section of the wavetable that is looped. To loop without a glitch (which can result in many undesirable harmonics), the value and slope at the start and end of the looping section will usually have to be matched. For some kinds of samples, this can be accomplished by putting the wavtable output through an appropriate low pass filter (using lpf), but it is even better to do the alignment and filter the result.
wavetable(name :: Symbol, amp, freq)Uses the named table. See register!.
Synth.maketable — Functionmaketable(L :: Int, f)Utility function to construct a table for use with wavetable. f is passed values in the range [0.0,1.0] to construct the table of the given length L.
Missing docstring for Synth.register!. Check Documenter's build log for details.