For this sort of mechanism, there are a number of things that might be desirable:
- That it can be extended easily to incorporate a larger number of input signals.
- That alternative fade-curves can be employed - linear cross-fading tends to leave a perceptual ‘hole in the middle’.
- ftmorf is quite nice but this limits the source signals to oscillators as opposed to any audio signal.
Here is what I came up with to address these ideas. It’s a little like Thrifleganger’s method but you only need to create a single crossfade function. There are four shapes suggested for cross-fading curves, I think Hanning sounds quite smooth.
<CsoundSynthesizer>
<CsOptions>
-dm0 -n
</CsOptions>
<CsInstruments>
sr = 44100
ksmps = 16
nchnls = 2
0dbfs = 1
instr 1
; create any number of signals...
a1 poscil 0.2,200
a2 poscil 0.2,400
a3 poscil 0.2,600
a4 poscil 0.2,800
a5 poscil 0.2,1000
; window shape used for fade-in/fade-out
iShape ftgen 0,0,4097,20,2,1 ; hanning
;iShape ftgen 0,0,4097,9,0.5,1,0 ; half-sine / equal-power
;iShape ftgen 0,0,4097,7,0,2048,1,2048,0 ; linear
;iShape ftgen 0,0,4097,20,6,1 ; gaussian
; crossfader. Should start at zero and can extend in integer steps for the number of signals to be mixed
kFade line 0, p3, 4 ; 0 = first signal only, 0.5 = mix of first two, 1 = second signal only etc...
; crossfade between any number of signals, note the sequence of table index offsets decreasing in integer steps
aOut sum a1 * tablei:k( (kFade + 1) *0.5 , iShape, 1), \
a2 * tablei:k( (kFade + 0) *0.5 , iShape, 1), \
a3 * tablei:k( (kFade - 1) *0.5 , iShape, 1), \
a4 * tablei:k( (kFade - 2) *0.5 , iShape, 1), \
a5 * tablei:k( (kFade - 3) *0.5 , iShape, 1)
outs aOut, aOut ; send audio to outputs
endin
</CsInstruments>
<CsScore>
i 1 0 7
</CsScore>
</CsoundSynthesizer>