Hi,
I'm working thru snd make-pvocoder, clm make-phase-vocoder trying to find more
info on the analyze, edit, and synthesize methods, such as the one below the
manual which shows a lambda func for the synthesis or resynthesis method. Most
of what I've been able to find is based on a gen from readin of a file or input
samples to access spectral data. I know this is a bit odd, but I was trying
to create designer spectra, by filling empty vectors with various numbers etc,
as opposed to getting the data from a readin generator. So far mine just
sounds like noise, so I don't have it quite right. I'm wondering if I'm on the
right track by using original vectors in the analyze, edit, and synthesize
methods. Or maybe I should just be creating sound files using something like
infinite sums, for example, and sticking with the readin generator to analyze
those sounds? I've looked at at the moving-spectrum and pins insts too.
Thank you,
Jim
(with-sound (:srate 22050 :statistics #t)
(let ((pv (make-phase-vocoder
(make-readin "oboe.snd")
512 4 128 1.0
#f ; no change to analysis method
#f ; no change to spectrum
(lambda (gen) ; resynthesis function
(float-vector-add! (phase-vocoder-amps gen)
(phase-vocoder-amp-increments gen))
(float-vector-add! (phase-vocoder-phase-increments gen)
(phase-vocoder-freqs gen))
(float-vector-add! (phase-vocoder-phases gen)
(phase-vocoder-phase-increments gen))
(let ((sum 0.0)
(n (length (phase-vocoder-amps gen
(do ((k 0 (+ k 1)))
((= k n))
(set! sum (+ sum (* (float-vector-ref (phase-vocoder-amps
gen) k)
(sin (* 0.5 (float-vector-ref
(phase-vocoder-phases gen) k)))
sum)
(do ((i 0 (+ i 1)))
((= i 44100))
(outa i (phase-vocoder pv)
(with-sound (:reverb nrev :srate 48000 :channels 2 :header-type mus-riff
:statistics #t :output "/myTest.wav" :play #t )
(let* ((samps (* 2 (mus-sound-framples "/Sample1.wav")))
(hop 512)
(jsampvec (make-float-vector hop)) ;create empty vectors
(jsampincvec (make-float-vector hop))
(jsfreqvec (make-float-vector hop))
(jsphasevec (make-float-vector hop))
(jsphaseincvec (make-float-vector hop))
(pv (make-phase-vocoder
(make-readin "/Sample1.wav")
512 4 128 1.75;fft-size, overlap, interp, pitch
#f ; no change to analysis
#f ; no change to edit
(lambda (gen) ; *attempt resynthesis function*
(do ((i 0 (+ i 1)))
((= i hop))
(set! (jsampvec i) (mus-random .02) ) ;fill vectors with
numbers
(set! (jsampincvec i) (mus-random .05) )
(set! (jsfreqvec i) (mus-random 500) )
(set! (jsphasevec i) (mus-random 360) )
(set! (jsphaseincvec i) (mus-random .02) )
) ;end do
(float-vector-add! jsampvec jsampincvec) ;add the vectors
(float-vector-add! jsphaseincvec jsfreqvec)
(float-vector-add! jsphasevec jsphaseincvec)
(let ((sum 0.0)
(n hop))
(do ((k 0 (+ k 1)))
((= k n))
(set! sum (+ sum (* (jsampvec k)(sin (* 0.5 (jsphasevec k)) ;create
waveform
)
sum)
)))
(do ((i 0 (+ i 1)))
((= i samps))
(outa i (* 4 (phase-vocoder pv)))
(outb i (* 4 (phase-vocoder pv)))
)))
___
Cmdist mailing list
Cmdist@ccrma.stanford.edu
https://cm-mail.stanford.edu/mailman/listinfo/cmdist
