aboutsummaryrefslogtreecommitdiff
path: root/externals/extra/0.43.2/fiddle~/fiddle~.c
diff options
context:
space:
mode:
Diffstat (limited to 'externals/extra/0.43.2/fiddle~/fiddle~.c')
-rw-r--r--externals/extra/0.43.2/fiddle~/fiddle~.c1851
1 files changed, 0 insertions, 1851 deletions
diff --git a/externals/extra/0.43.2/fiddle~/fiddle~.c b/externals/extra/0.43.2/fiddle~/fiddle~.c
deleted file mode 100644
index ea676784..00000000
--- a/externals/extra/0.43.2/fiddle~/fiddle~.c
+++ /dev/null
@@ -1,1851 +0,0 @@
-/* Copyright (c) 1997-1999 Miller Puckette and Ted Apel.
-* For information on usage and redistribution, and for a DISCLAIMER OF ALL
-* WARRANTIES, see the file, "LICENSE.txt," in this distribution. */
-
-/*
- * Fiddle is a pitch tracker hardwired to have hop size ("H") equal to
- * half its window size ("N").
- *
- * This version should compile for Max "0.26," JMAX, Pd, or Max/MSP.
- *
- * The "lastanalysis" field holds the shifted FT of the previous H
- * samples. The buffer contains in effect points 1/2, 3/2, ..., (N-1)/2
- * of the DTFT of a real vector of length N, half of whose points are zero,
- * i.e., only the first H points are used. Put another way, we get the
- * the odd-numbered points of the FFT of the H points, zero padded to 4*H in
- * length. The integer points 0, 1, ..., H-1
- * are found by interpolating these others, using the fact that the
- * half-integer points are band-limited (they only have positive frequencies.)
- * To facilitate the interpolation the "lastanalysis" buffer contains
- * FILTSIZE extra points (1/2-FILTSIZE, ..., -1/2) at the beginning and
- * FILTSIZE again at the end ((N+1)/2, ..., FILTSIZE+(N-1)/2). The buffer
- * therefore has N+4*FILTSIZE floating-point numbers in it.
- *
- * after doing this I found out that you can just do a real FFT
- * of the H new points, zero-padded to contain N points, and using a similar
- * but simpler interpolation scheme you can still get 2N points of the DTFT
- * of the N points. Jean Laroche is a big fat hen.
- *
- */
-
-
-/* These pragmas are only used for MSVC, not MinGW or Cygwin <hans@at.or.at> */
-#ifdef _MSC_VER
-#pragma warning (disable: 4305 4244)
-#endif
-
-/* this #ifdef does nothing, but its there... */
-#ifdef _WIN32
-#define flog log
-#define fexp exp
-#define fsqrt sqrt
-#else
-#define flog log
-#define fexp exp
-#define fsqrt sqrt
-#endif
-
-char fiddle_version[] = "fiddle version 1.1 TEST4";
-
-#ifdef JMAX
-#include "fts.h"
-#include <stdio.h>
-#include <stdlib.h>
-typedef float t_float;
-typedef float t_floatarg;
-typedef fts_symbol_t t_symbol;
-
-static void *getbytes(size_t nbytes)
-{
- void *ret;
- if (nbytes < 1) nbytes = 1;
- ret = (void *)malloc(nbytes);
- return (ret);
-}
-
-static void *resizebytes(void *old, size_t oldsize, size_t newsize)
-{
- void *ret;
- if (newsize < 1) newsize = 1;
- ret = (void *)realloc((char *)old, newsize);
- return (ret);
-}
-
-static void freebytes(void *fatso, size_t nbytes)
-{
- free(fatso);
-}
-
-#define CLASSNAME "fiddle"
-
-#define OUTLETpower 5
-#define OUTLETmicropitch1 4
-#define OUTLETmicropitch2 3
-#define OUTLETmicropitch3 2
-#define OUTLETattack 1
-#define OUTLETpitch 0
-
-static fts_symbol_t *dsp_symbol = 0;
-#define error post
-
-#endif /* FTS */
-
-#ifdef MAX26
-#define t_floatarg double
-#include "m_extern.h"
-#include "d_graph.h"
-#include "d_ugen.h"
-#endif /* MAX26 */
-
-#ifdef PD
-#include "m_pd.h"
-#endif /* PD */
-
-#ifdef MSP
-#define flog log
-#define fexp exp
-#define fsqrt sqrt
-#endif /* MSP */
-
-#ifdef MSP
-#define t_floatarg double
-#include "ext.h"
-#include "z_dsp.h"
-#include "fft_mayer.proto.h"
-
-#endif /* MSP */
-
-#include <math.h>
-
-
-#define MINBIN 3
-#define DEFAMPLO 40
-#define DEFAMPHI 50
-#define DEFATTACKTIME 100
-#define DEFATTACKTHRESH 10
-#define DEFVIBTIME 50
-#define DEFVIBDEPTH 0.5
-#define GLISS 0.7f
-#define DBFUDGE 30.8f
-#define MINFREQINBINS 5 /* minimum frequency in bins for reliable output */
-
-#define MAXNPITCH 3
-#define MAXHIST 3 /* find N hottest peaks in histogram */
-
-#define MAXPOINTS 8192
-#define MINPOINTS 128
-#define DEFAULTPOINTS 1024
-
-#define HISTORY 20
-#define MAXPEAK 100 /* maximum number of peaks */
-#define DEFNPEAK 20 /* default number of peaks */
-
-#define MAXNPEAK (MAXLOWPEAK + MAXSTRONGPEAK)
-#define MINBW (0.03f) /* consider BW >= 0.03 FFT bins */
-
-#define BINPEROCT 48 /* bins per octave */
-#define BPERO_OVER_LOG2 69.24936196f /* BINSPEROCT/log(2) */
-#define FACTORTOBINS (t_float)(4/0.0145453) /* 4 / (pow(2.,1/48.) - 1) */
-#define BINGUARD 10 /* extra bins to throw in front */
-#define PARTIALDEVIANCE 0.023f /* acceptable partial detuning in % */
-#define LOGTODB 4.34294481903f /* 20/log(10) */
-
-#define KNOCKTHRESH 10.f /* don't know how to describe this */
-
-
-static t_float sigfiddle_partialonset[] =
-{
-0,
-48,
-76.0782000346154967102,
-96,
-111.45254855459339269887,
-124.07820003461549671089,
-134.75303625876499715823,
-144,
-152.15640006923099342109,
-159.45254855459339269887,
-166.05271769459026829915,
-172.07820003461549671088,
-177.62110647077242370064,
-182.75303625876499715892,
-187.53074858920888940907,
-192,
-};
-
-#define NPARTIALONSET ((int)(sizeof(sigfiddle_partialonset)/sizeof(t_float)))
-
-static int sigfiddle_intpartialonset[] =
-{
-0,
-48,
-76,
-96,
-111,
-124,
-135,
-144,
-152,
-159,
-166,
-172,
-178,
-183,
-188,
-192,
-};
-
-/* these coefficients, which come from the "upsamp" subdirectory,
-are a filter kernel for upsampling by a factor of two, assuming
-the sound to be upsampled has no energy above half the Nyquist, i.e.,
-that it's already 2x oversampled compared to the theoretically possible
-sample rate. I got these by trial and error. */
-
-#define FILT1 ((t_float)(.5 * 1.227054))
-#define FILT2 ((t_float)(.5 * -0.302385))
-#define FILT3 ((t_float)(.5 * 0.095326))
-#define FILT4 ((t_float)(.5 * -0.022748))
-#define FILT5 ((t_float)(.5 * 0.002533))
-#define FILTSIZE 5
-
-typedef struct peakout /* a peak for output */
-{
- t_float po_freq; /* frequency in hz */
- t_float po_amp; /* amplitude */
-} t_peakout;
-
-typedef struct peak /* a peak for analysis */
-{
- t_float p_freq; /* frequency in bins */
- t_float p_width; /* peak width in bins */
- t_float p_pow; /* peak power */
- t_float p_loudness; /* 4th root of power */
- t_float *p_fp; /* pointer back to spectrum */
-} t_peak;
-
-typedef struct histopeak
-{
- t_float h_pitch; /* estimated pitch */
- t_float h_value; /* value of peak */
- t_float h_loud; /* combined strength of found partials */
- int h_index; /* index of bin holding peak */
- int h_used; /* true if an x_hist entry points here */
-} t_histopeak;
-
-typedef struct pitchhist /* struct for keeping history by pitch */
-{
- t_float h_pitch; /* pitch to output */
- t_float h_amps[HISTORY]; /* past amplitudes */
- t_float h_pitches[HISTORY]; /* past pitches */
- t_float h_noted; /* last pitch output */
- int h_age; /* number of frames pitch has been there */
- t_histopeak *h_wherefrom; /* new histogram peak to incorporate */
- void *h_outlet;
-} t_pitchhist;
-
-typedef struct sigfiddle /* instance struct */
-{
-#ifdef JMAX
- fts_object_t x_h; /* object header */
- fts_alarm_t x_clock; /* callback for timeouts */
-#endif
-#ifdef MAX26
- t_head x_h; /* header for tilde objects */
- t_sig *x_io[IN1+OUT0]; /* number of signal inputs and outputs */
- void *x_clock; /* a "clock" object */
-#endif
-#ifdef PD
- t_object x_ob; /* object header */
- t_clock *x_clock; /* callback for timeouts */
-#endif
-#ifdef MSP
- t_pxobject x_obj;
- void *x_clock;
- long x_downsample; /* downsample feature because of
- MSP's large sig vector sizes */
-#endif
- t_float *x_inbuf; /* buffer to analyze, npoints/2 elems */
- t_float *x_lastanalysis; /* FT of last buffer (see main comment) */
- t_float *x_spiral; /* 1/4-wave complex exponential */
- t_peakout *x_peakbuf; /* spectral peaks for output */
- int x_npeakout; /* number of spectral peaks to output */
- int x_npeakanal; /* number of spectral peaks to analyze */
- int x_phase; /* number of points since last output */
- int x_histphase; /* phase into amplitude history vector */
- int x_hop; /* period of output, npoints/2 */
- t_float x_sr; /* sample rate */
- t_pitchhist x_hist[MAXNPITCH]; /* history of current pitches */
- int x_nprint; /* how many periods to print */
- int x_npitch; /* number of simultaneous pitches */
- t_float x_dbs[HISTORY]; /* DB history, indexed by "histphase" */
- t_float x_peaked; /* peak since last attack */
- int x_dbage; /* number of bins DB has met threshold */
- int x_auto; /* true if generating continuous output */
-/* parameters */
- t_float x_amplo;
- t_float x_amphi;
- int x_attacktime;
- int x_attackbins;
- t_float x_attackthresh;
- int x_vibtime;
- int x_vibbins;
- t_float x_vibdepth;
- t_float x_npartial;
-/* outlets & clock */
- void *x_envout;
- int x_attackvalue;
- void *x_attackout;
- void *x_noteout;
- void *x_peakout;
-} t_sigfiddle;
-
-#if CHECKER
-t_float fiddle_checker[1024];
-#endif
-
-#ifdef MSP
-/* Mac compiler requires prototypes for everything */
-
-int sigfiddle_ilog2(int n);
-t_float fiddle_mtof(t_float f);
-t_float fiddle_ftom(t_float f);
-void sigfiddle_doit(t_sigfiddle *x);
-void sigfiddle_debug(t_sigfiddle *x);
-void sigfiddle_print(t_sigfiddle *x);
-void sigfiddle_assist(t_sigfiddle *x, void *b, long m, long a, char *s);
-void sigfiddle_amprange(t_sigfiddle *x, double amplo, double amphi);
-void sigfiddle_reattack(t_sigfiddle *x, t_floatarg attacktime, t_floatarg
-attackthresh);
-void sigfiddle_vibrato(t_sigfiddle *x, t_floatarg vibtime, t_floatarg
-vibdepth);
-void sigfiddle_npartial(t_sigfiddle *x, double npartial);
-void sigfiddle_auto(t_sigfiddle *x, t_floatarg f);
-void sigfiddle_setnpoints(t_sigfiddle *x, t_floatarg f);
-int sigfiddle_doinit(t_sigfiddle *x, long npoints, long npitch, long
-npeakanal, long npeakout);
-static t_int *fiddle_perform(t_int *w);
-void sigfiddle_dsp(t_sigfiddle *x, t_signal **sp);
-void sigfiddle_tick(t_sigfiddle *x);
-void sigfiddle_bang(t_sigfiddle *x);
-void sigfiddle_ff(t_sigfiddle *x);
-void *sigfiddle_new(long npoints, long npitch,
- long npeakanal, long npeakout);
-void msp_fft(t_float *buf, long np, long inv);
-t_float msp_ffttemp[MAXPOINTS*2];
-int errno;
-#endif
-
-int sigfiddle_ilog2(int n)
-{
- int ret = -1;
- while (n)
- {
- n >>= 1;
- ret++;
- }
- return (ret);
-}
-
-t_float fiddle_mtof(t_float f)
-{
- return (8.17579891564 * exp(.0577622650 * f));
-}
-
-t_float fiddle_ftom(t_float f)
-{
- return (17.3123405046 * log(.12231220585 * f));
-}
-#define ftom fiddle_ftom
-#define mtof fiddle_mtof
-
-void sigfiddle_doit(t_sigfiddle *x)
-{
-#ifdef MSP
- /* prevents interrupt-level stack overflow crash with Netscape. */
- static t_float spect1[4*MAXPOINTS];
- static t_float spect2[MAXPOINTS + 4*FILTSIZE];
-#else
- t_float spect1[4*MAXPOINTS];
- t_float spect2[MAXPOINTS + 4*FILTSIZE];
-#endif
-#if CHECKER
- t_float checker3[4*MAXPOINTS];
-#endif
-
- t_peak peaklist[MAXPEAK + 1], *pk1;
- t_peakout *pk2;
- t_histopeak histvec[MAXHIST], *hp1;
- int i, j, k, hop = x->x_hop, n = 2*hop, npeak, npitch,
- logn = sigfiddle_ilog2(n), newphase, oldphase;
- t_float *fp, *fp1, *fp2, *fp3, total_power, total_loudness, total_db;
- t_float maxbin = BINPEROCT * (logn-2), *histogram = spect2 + BINGUARD;
- t_pitchhist *phist;
- t_float hzperbin = x->x_sr / (2.0f * n);
- int npeakout = x->x_npeakout, npeakanal = x->x_npeakanal;
- int npeaktot = (npeakout > npeakanal ? npeakout : npeakanal);
-
- oldphase = x->x_histphase;
- newphase = x->x_histphase + 1;
- if (newphase == HISTORY) newphase = 0;
- x->x_histphase = newphase;
-
- /*
- * multiply the H points by a 1/4-wave complex exponential,
- * and take FFT of the result.
- */
- for (i = 0, fp1 = x->x_inbuf, fp2 = x->x_spiral, fp3 = spect1;
- i < hop; i++, fp1++, fp2 += 2, fp3 += 2)
- fp3[0] = fp1[0] * fp2[0], fp3[1] = fp1[0] * fp2[1];
-
-#ifdef MAX26
- fft(spect1, hop, 0);
-#endif
-#ifdef PD
- pd_fft(spect1, hop, 0);
-#endif
-#ifdef JMAX
- fts_cfft_inplc((complex *)spect1, hop);
-#endif
-#ifdef MSP
- msp_fft(spect1,hop,0);
-#endif
- /*
- * now redistribute the points to get in effect the odd-numbered
- * points of the FFT of the H points, zero padded to 4*H in length.
- */
- for (i = 0, fp1 = spect1, fp2 = spect2 + (2*FILTSIZE);
- i < (hop>>1); i++, fp1 += 2, fp2 += 4)
- fp2[0] = fp1[0], fp2[1] = fp1[1];
- for (i = 0, fp1 = spect1 + n - 2, fp2 = spect2 + (2*FILTSIZE+2);
- i < (hop>>1); i++, fp1 -= 2, fp2 += 4)
- fp2[0] = fp1[0], fp2[1] = -fp1[1];
- for (i = 0, fp1 = spect2 + (2*FILTSIZE), fp2 = spect2 + (2*FILTSIZE-2);
- i<FILTSIZE; i++, fp1+=2, fp2-=2)
- fp2[0] = fp1[0], fp2[1] = -fp1[1];
- for (i = 0, fp1 = spect2 + (2*FILTSIZE+n-2), fp2 = spect2 + (2*FILTSIZE+n);
- i<FILTSIZE; i++, fp1-=2, fp2+=2)
- fp2[0] = fp1[0], fp2[1] = -fp1[1];
-#if 0
- {
- fp = spect2 + 2*FILTSIZE;
- post("x1 re %12.4f %12.4f %12.4f %12.4f %12.4f",
- fp[0], fp[2], fp[4], fp[6], fp[8]);
- post("x1 im %12.4f %12.4f %12.4f %12.4f %12.4f",
- fp[1], fp[3], fp[5], fp[7], fp[9]);
- }
-#endif
- /* spect2 is now prepared; now combine spect2 and lastanalysis into
- * spect1. Odd-numbered points of spect1 are the points of "last"
- * plus (-i, i, -i, ...) times spect1. Even-numbered points are
- * the interpolated points of "last" plus (1, -1, 1, ...) times the
- * interpolated points of spect1.
- *
- * To interpolate, take FILT1 exp(-pi/4) times
- * the previous point, FILT2*exp(-3*pi/4) times 3 bins before,
- * etc, and FILT1 exp(pi/4), FILT2 exp(3pi/4), etc., to weight
- * the +1, +3, etc., points.
- *
- * In this calculation, we take (1, i, -1, -i, 1) times the
- * -9, -7, ..., -1 points, and (i, -1, -i, 1, i) times the 1, 3,..., 9
- * points of the OLD spectrum, alternately adding and subtracting
- * the new spectrum to the old; then we multiply the whole thing
- * by exp(-i pi/4).
- */
- for (i = 0, fp1 = spect1, fp2 = x->x_lastanalysis + 2*FILTSIZE,
- fp3 = spect2 + 2*FILTSIZE;
- i < (hop>>1); i++)
- {
- t_float re, im;
-
- re= FILT1 * ( fp2[ -2] -fp2[ 1] +fp3[ -2] -fp3[ 1]) +
- FILT2 * ( fp2[ -3] -fp2[ 2] +fp3[ -3] -fp3[ 2]) +
- FILT3 * (-fp2[ -6] +fp2[ 5] -fp3[ -6] +fp3[ 5]) +
- FILT4 * (-fp2[ -7] +fp2[ 6] -fp3[ -7] +fp3[ 6]) +
- FILT5 * ( fp2[-10] -fp2[ 9] +fp3[-10] -fp3[ 9]);
-
- im= FILT1 * ( fp2[ -1] +fp2[ 0] +fp3[ -1] +fp3[ 0]) +
- FILT2 * (-fp2[ -4] -fp2[ 3] -fp3[ -4] -fp3[ 3]) +
- FILT3 * (-fp2[ -5] -fp2[ 4] -fp3[ -5] -fp3[ 4]) +
- FILT4 * ( fp2[ -8] +fp2[ 7] +fp3[ -8] +fp3[ 7]) +
- FILT5 * ( fp2[ -9] +fp2[ 8] +fp3[ -9] +fp3[ 8]);
-
- fp1[0] = 0.7071f * (re + im);
- fp1[1] = 0.7071f * (im - re);
- fp1[4] = fp2[0] + fp3[1];
- fp1[5] = fp2[1] - fp3[0];
-
- fp1 += 8, fp2 += 2, fp3 += 2;
- re= FILT1 * ( fp2[ -2] -fp2[ 1] -fp3[ -2] +fp3[ 1]) +
- FILT2 * ( fp2[ -3] -fp2[ 2] -fp3[ -3] +fp3[ 2]) +
- FILT3 * (-fp2[ -6] +fp2[ 5] +fp3[ -6] -fp3[ 5]) +
- FILT4 * (-fp2[ -7] +fp2[ 6] +fp3[ -7] -fp3[ 6]) +
- FILT5 * ( fp2[-10] -fp2[ 9] -fp3[-10] +fp3[ 9]);
-
- im= FILT1 * ( fp2[ -1] +fp2[ 0] -fp3[ -1] -fp3[ 0]) +
- FILT2 * (-fp2[ -4] -fp2[ 3] +fp3[ -4] +fp3[ 3]) +
- FILT3 * (-fp2[ -5] -fp2[ 4] +fp3[ -5] +fp3[ 4]) +
- FILT4 * ( fp2[ -8] +fp2[ 7] -fp3[ -8] -fp3[ 7]) +
- FILT5 * ( fp2[ -9] +fp2[ 8] -fp3[ -9] -fp3[ 8]);
-
- fp1[0] = 0.7071f * (re + im);
- fp1[1] = 0.7071f * (im - re);
- fp1[4] = fp2[0] - fp3[1];
- fp1[5] = fp2[1] + fp3[0];
-
- fp1 += 8, fp2 += 2, fp3 += 2;
- }
-#if 0
- if (x->x_nprint)
- {
- for (i = 0, fp = spect1; i < 16; i++, fp+= 4)
- post("spect %d %f %f --> %f", i, fp[0], fp[1],
- sqrt(fp[0] * fp[0] + fp[1] * fp[1]));
- }
-#endif
- /* copy new spectrum out */
- for (i = 0, fp1 = spect2, fp2 = x->x_lastanalysis;
- i < n + 4*FILTSIZE; i++) *fp2++ = *fp1++;
-
- for (i = 0; i < MINBIN; i++) spect1[4*i + 2] = spect1[4*i + 3] = 0;
- /* starting at bin MINBIN, compute hanning windowed power spectrum */
- for (i = MINBIN, fp1 = spect1+4*MINBIN, total_power = 0;
- i < n-2; i++, fp1 += 4)
- {
- t_float re = fp1[0] - 0.5 * (fp1[-8] + fp1[8]);
- t_float im = fp1[1] - 0.5 * (fp1[-7] + fp1[9]);
- fp1[3] = (total_power += (fp1[2] = re * re + im * im));
- }
-
- if (total_power > 1e-9f)
- {
- total_db = (100.f - DBFUDGE) + LOGTODB * log(total_power/n);
- total_loudness = fsqrt(fsqrt(total_power));
- if (total_db < 0) total_db = 0;
- }
- else total_db = total_loudness = 0;
- /* store new db in history vector */
- x->x_dbs[newphase] = total_db;
- if (total_db < x->x_amplo) goto nopow;
-#if 1
- if (x->x_nprint) post("power %f", total_power);
-#endif
-
-#if CHECKER
- /* verify that our FFT resampling thing is putting out good results */
- for (i = 0; i < hop; i++)
- {
- checker3[2*i] = fiddle_checker[i];
- checker3[2*i + 1] = 0;
- checker3[n + 2*i] = fiddle_checker[i] = x->x_inbuf[i];
- checker3[n + 2*i + 1] = 0;
- }
- for (i = 2*n; i < 4*n; i++) checker3[i] = 0;
- fft(checker3, 2*n, 0);
- if (x->x_nprint)
- {
- for (i = 0, fp = checker3; i < 16; i++, fp += 2)
- post("spect %d %f %f --> %f", i, fp[0], fp[1],
- sqrt(fp[0] * fp[0] + fp[1] * fp[1]));
- }
-
-#endif
- npeak = 0;
-
- /* search for peaks */
- for (i = MINBIN, fp = spect1+4*MINBIN, pk1 = peaklist;
- i < n-2 && npeak < npeaktot; i++, fp += 4)
- {
- t_float height = fp[2], h1 = fp[-2], h2 = fp[6];
- t_float totalfreq, pfreq, f1, f2, m, var, stdev;
-
- if (height < h1 || height < h2 ||
- h1 < 0.00001f*total_power || h2 < 0.00001f*total_power)
- continue;
-
- /* use an informal phase vocoder to estimate the frequency.
- Do this for the two adjacent bins too. */
- pfreq= ((fp[-8] - fp[8]) * (2.0f * fp[0] - fp[8] - fp[-8]) +
- (fp[-7] - fp[9]) * (2.0f * fp[1] - fp[9] - fp[-7])) /
- (2.0f * height);
- f1= ((fp[-12] - fp[4]) * (2.0f * fp[-4] - fp[4] - fp[-12]) +
- (fp[-11] - fp[5]) * (2.0f * fp[-3] - fp[5] - fp[-11])) /
- (2.0f * h1) - 1;
- f2= ((fp[-4] - fp[12]) * (2.0f * fp[4] - fp[12] - fp[-4]) +
- (fp[-3] - fp[13]) * (2.0f * fp[5] - fp[13] - fp[-3])) /
- (2.0f * h2) + 1;
-
- /* get sample mean and variance of the three */
- m = 0.333333f * (pfreq + f1 + f2);
- var = 0.5f * ((pfreq-m)*(pfreq-m) + (f1-m)*(f1-m) + (f2-m)*(f2-m));
-
- totalfreq = i + m;
- if (var * total_power > KNOCKTHRESH * height || var < 1e-30)
- {
-#if 0
- if (x->x_nprint)
- post("cancel: %.2f hz, index %.1f, power %.5f, stdev=%.2f",
- totalfreq * hzperbin, BPERO_OVER_LOG2 * log(totalfreq) - 96,
- height, sqrt(var));
-#endif
- continue;
- }
- stdev = fsqrt(var);
- if (totalfreq < 4)
- {
- if (x->x_nprint) post("oops: was %d, freq %f, m %f, stdev %f h %f",
- i, totalfreq, m, stdev, height);
- totalfreq = 4;
- }
- pk1->p_width = stdev;
-
- pk1->p_pow = height;
- pk1->p_loudness = fsqrt(fsqrt(height));
- pk1->p_fp = fp;
- pk1->p_freq = totalfreq;
- npeak++;
-#if 1
- if (x->x_nprint)
- {
- post("peak: %.2f hz. index %.1f, power %.5f, stdev=%.2f",
- pk1->p_freq * hzperbin,
- BPERO_OVER_LOG2 * log(pk1->p_freq) - 96,
- height, stdev);
- }
-#endif
- pk1++;
- }
-
- /* prepare the raw peaks for output */
- for (i = 0, pk1 = peaklist, pk2 = x->x_peakbuf; i < npeak;
- i++, pk1++, pk2++)
- {
- t_float loudness = pk1->p_loudness;
- if (i >= npeakout) break;
- pk2->po_freq = hzperbin * pk1->p_freq;
- pk2->po_amp = (2. / (t_float)n) * (loudness * loudness);
- }
- for (; i < npeakout; i++, pk2++) pk2->po_amp = pk2->po_freq = 0;
-
- /* now, working back into spect2, make a sort of "liklihood"
- * spectrum. Proceeding in 48ths of an octave, from 2 to
- * n/2 (in bins), the likelihood of each pitch range is contributed
- * to by every peak in peaklist that's an integer multiple of it
- * in frequency.
- */
-
- if (npeak > npeakanal) npeak = npeakanal; /* max # peaks to analyze */
- for (i = 0, fp1 = histogram; i < maxbin; i++) *fp1++ = 0;
- for (i = 0, pk1 = peaklist; i < npeak; i++, pk1++)
- {
- t_float pit = BPERO_OVER_LOG2 * flog(pk1->p_freq) - 96.0;
- t_float binbandwidth = FACTORTOBINS * pk1->p_width/pk1->p_freq;
- t_float putbandwidth = (binbandwidth < 2 ? 2 : binbandwidth);
- t_float weightbandwidth = (binbandwidth < 1.0 ? 1.0 : binbandwidth);
- /* t_float weightamp = 1.0f + 3.0f * pk1->p_pow / pow; */
- t_float weightamp = 4. * pk1->p_loudness / total_loudness;
- for (j = 0, fp2 = sigfiddle_partialonset; j < NPARTIALONSET; j++, fp2++)
- {
- t_float bin = pit - *fp2;
- if (bin < maxbin)
- {
- t_float para, pphase, score = 30.0 * weightamp /
- ((j+x->x_npartial) * weightbandwidth);
- int firstbin = bin + 0.5f - 0.5f * putbandwidth;
- int lastbin = bin + 0.5f + 0.5f * putbandwidth;
- int ibw = lastbin - firstbin;
- if (firstbin < -BINGUARD) break;
- para = 1.0f / (putbandwidth * putbandwidth);
- for (k = 0, fp3 = histogram + firstbin,
- pphase = firstbin-bin; k <= ibw;
- k++, fp3++, pphase += 1.0f)
- {
- *fp3 += score * (1.0f - para * pphase * pphase);
- }
- }
- }
- }
-#if 1
- if (x->x_nprint)
- {
- for (i = 0; i < 6*5; i++)
- {
- t_float fhz = hzperbin * exp ((8*i + 96) * (1./BPERO_OVER_LOG2));
- if (!(i % 6)) post("-- bin %d pitch %f freq %f----", 8*i,
- ftom(fhz), fhz);;
- post("%3d %3d %3d %3d %3d %3d %3d %3d",
- (int)(histogram[8*i]),
- (int)(histogram[8*i+1]),
- (int)(histogram[8*i+2]),
- (int)(histogram[8*i+3]),
- (int)(histogram[8*i+4]),
- (int)(histogram[8*i+5]),
- (int)(histogram[8*i+6]),
- (int)(histogram[8*i+7]));
- }
- }
-
-#endif
-
- /*
- * Next we find up to NPITCH strongest peaks in the histogram.
- * if a peak is related to a stronger one via an interval in
- * the sigfiddle_partialonset array, we suppress it.
- */
-
- for (npitch = 0; npitch < x->x_npitch; npitch++)
- {
- int indx;
- t_float best;
- if (npitch)
- {
- for (best = 0, indx = -1, j=1; j < maxbin-1; j++)
- {
- if (histogram[j] > best && histogram[j] > histogram[j-1] &&
- histogram[j] > histogram[j+1])
- {
- for (k = 0; k < npitch; k++)
- if (histvec[k].h_index == j)
- goto peaknogood;
- for (k = 0; k < NPARTIALONSET; k++)
- {
- if (j - sigfiddle_intpartialonset[k] < 0) break;
- if (histogram[j - sigfiddle_intpartialonset[k]]
- > histogram[j]) goto peaknogood;
- }
- for (k = 0; k < NPARTIALONSET; k++)
- {
- if (j + sigfiddle_intpartialonset[k] >= maxbin) break;
- if (histogram[j + sigfiddle_intpartialonset[k]]
- > histogram[j]) goto peaknogood;
- }
- indx = j;
- best = histogram[j];
- }
- peaknogood: ;
- }
- }
- else
- {
- for (best = 0, indx = -1, j=0; j < maxbin; j++)
- if (histogram[j] > best)
- indx = j, best = histogram[j];
- }
- if (indx < 0) break;
- histvec[npitch].h_value = best;
- histvec[npitch].h_index = indx;
- }
-#if 1
- if (x->x_nprint)
- {
- for (i = 0; i < npitch; i++)
- {
- post("index %d freq %f --> value %f", histvec[i].h_index,
- exp((1./BPERO_OVER_LOG2) * (histvec[i].h_index + 96)),
- histvec[i].h_value);
- post("next %f , prev %f",
- exp((1./BPERO_OVER_LOG2) * (histvec[i].h_index + 97)),
- exp((1./BPERO_OVER_LOG2) * (histvec[i].h_index + 95)) );
- }
- }
-#endif
-
- /* for each histogram peak, we now search back through the
- * FFT peaks. A peak is a pitch if either there are several
- * harmonics that match it, or else if (a) the fundamental is
- * present, and (b) the sum of the powers of the contributing peaks
- * is at least 1/100 of the total power.
- *
- * A peak is a contributor if its frequency is within 25 cents of
- * a partial from 1 to 16.
- *
- * Finally, we have to be at least 5 bins in frequency, which
- * corresponds to 2-1/5 periods fitting in the analysis window.
- */
-
- for (i = 0; i < npitch; i++)
- {
- t_float cumpow = 0, cumstrength = 0, freqnum = 0, freqden = 0;
- int npartials = 0, nbelow8 = 0;
- /* guessed-at frequency in bins */
- t_float putfreq = fexp((1.0 / BPERO_OVER_LOG2) *
- (histvec[i].h_index + 96.0f));
- for (j = 0; j < npeak; j++)
- {
- t_float fpnum = peaklist[j].p_freq/putfreq;
- int pnum = fpnum + 0.5f;
- t_float fipnum = pnum;
- t_float deviation;
- if (pnum > 16 || pnum < 1) continue;
- deviation = 1.0f - fpnum/fipnum;
- if (deviation > -PARTIALDEVIANCE && deviation < PARTIALDEVIANCE)
- {
- /*
- * we figure this is a partial since it's within 1/4 of
- * a halftone of a multiple of the putative frequency.
- */
-
- t_float stdev, weight;
- npartials++;
- if (pnum < 8) nbelow8++;
- cumpow += peaklist[j].p_pow;
- cumstrength += fsqrt(fsqrt(peaklist[j].p_pow));
- stdev = (peaklist[j].p_width > MINBW ?
- peaklist[j].p_width : MINBW);
- weight = 1.0f / ((stdev*fipnum) * (stdev*fipnum));
- freqden += weight;
- freqnum += weight * peaklist[j].p_freq/fipnum;
-#if 1
- if (x->x_nprint)
- {
- post("peak %d partial %d f=%f w=%f",
- j, pnum, peaklist[j].p_freq/fipnum, weight);
- }
-#endif
- }
-#if 1
- else if (x->x_nprint) post("peak %d partial %d dev %f",
- j, pnum, deviation);
-#endif
- }
- if ((nbelow8 < 4 || npartials < 7) && cumpow < 0.01f * total_power)
- histvec[i].h_value = 0;
- else
- {
- t_float pitchpow = (cumstrength * cumstrength) *
- (cumstrength * cumstrength);
- t_float freqinbins = freqnum/freqden;
- /* check for minimum output frequency */
-
- if (freqinbins < MINFREQINBINS)
- histvec[i].h_value = 0;
- else
- {
- /* we passed all tests... save the values we got */
- histvec[i].h_pitch = ftom(hzperbin * freqnum/freqden);
- histvec[i].h_loud = (100.0f -DBFUDGE) +
- (LOGTODB) * log(pitchpow/n);
- }
- }
- }
-#if 1
- if (x->x_nprint)
- {
- for (i = 0; i < npitch; i++)
- {
- if (histvec[i].h_value > 0)
- post("index %d pit %f loud %f", histvec[i].h_index,
- histvec[i].h_pitch, histvec[i].h_loud);
- else post("-- cancelled --");
- }
- }
-#endif
-
- /* now try to find continuous pitch tracks that match the new
- * pitches. First mark each peak unmatched.
- */
- for (i = 0, hp1 = histvec; i < npitch; i++, hp1++)
- hp1->h_used = 0;
-
- /* for each old pitch, try to match a new one to it. */
- for (i = 0, phist = x->x_hist; i < x->x_npitch; i++, phist++)
- {
- t_float thispitch = phist->h_pitches[oldphase];
- phist->h_pitch = 0; /* no output, thanks */
- phist->h_wherefrom = 0;
- if (thispitch == 0.0f) continue;
- for (j = 0, hp1 = histvec; j < npitch; j++, hp1++)
- if ((hp1->h_value > 0) && hp1->h_pitch > thispitch - GLISS
- && hp1->h_pitch < thispitch + GLISS)
- {
- phist->h_wherefrom = hp1;
- hp1->h_used = 1;
- }
- }
- for (i = 0, hp1 = histvec; i < npitch; i++, hp1++)
- if ((hp1->h_value > 0) && !hp1->h_used)
- {
- for (j = 0, phist = x->x_hist; j < x->x_npitch; j++, phist++)
- if (!phist->h_wherefrom)
- {
- phist->h_wherefrom = hp1;
- phist->h_age = 0;
- phist->h_noted = 0;
- hp1->h_used = 1;
- goto happy;
- }
- break;
- happy: ;
- }
- /* copy the pitch info into the history vector */
- for (i = 0, phist = x->x_hist; i < x->x_npitch; i++, phist++)
- {
- if (phist->h_wherefrom)
- {
- phist->h_amps[newphase] = phist->h_wherefrom->h_loud;
- phist->h_pitches[newphase] =
- phist->h_wherefrom->h_pitch;
- (phist->h_age)++;
- }
- else
- {
- phist->h_age = 0;
- phist->h_amps[newphase] = phist->h_pitches[newphase] = 0;
- }
- }
-#if 1
- if (x->x_nprint)
- {
- post("vibrato %d %f", x->x_vibbins, x->x_vibdepth);
- for (i = 0, phist = x->x_hist; i < x->x_npitch; i++, phist++)
- {
- post("noted %f, age %d", phist->h_noted, phist->h_age);
-#ifndef I860
- post("values %f %f %f %f %f",
- phist->h_pitches[newphase],
- phist->h_pitches[(newphase + HISTORY-1)%HISTORY],
- phist->h_pitches[(newphase + HISTORY-2)%HISTORY],
- phist->h_pitches[(newphase + HISTORY-3)%HISTORY],
- phist->h_pitches[(newphase + HISTORY-4)%HISTORY]);
-#endif
- }
- }
-#endif
- /* look for envelope attacks */
-
- x->x_attackvalue = 0;
-
- if (x->x_peaked)
- {
- if (total_db > x->x_amphi)
- {
- int binlook = newphase - x->x_attackbins;
- if (binlook < 0) binlook += HISTORY;
- if (total_db > x->x_dbs[binlook] + x->x_attackthresh)
- {
- x->x_attackvalue = 1;
- x->x_peaked = 0;
- }
- }
- }
- else
- {
- int binlook = newphase - x->x_attackbins;
- if (binlook < 0) binlook += HISTORY;
- if (x->x_dbs[binlook] > x->x_amphi && x->x_dbs[binlook] > total_db)
- x->x_peaked = 1;
- }
-
- /* for each current frequency track, test for a new note using a
- * stability criterion. Later perhaps we should also do as in
- * pitch~ and check for unstable notes a posteriori when
- * there's a new attack with no note found since the last onset;
- * but what's an attack &/or onset when we're polyphonic?
- */
-
- for (i = 0, phist = x->x_hist; i < x->x_npitch; i++, phist++)
- {
- /*
- * if we've found a pitch but we've now strayed from it turn
- * it off.
- */
- if (phist->h_noted)
- {
- if (phist->h_pitches[newphase] > phist->h_noted + x->x_vibdepth
- || phist->h_pitches[newphase] < phist->h_noted - x->x_vibdepth)
- phist->h_noted = 0;
- }
- else
- {
- if (phist->h_wherefrom && phist->h_age >= x->x_vibbins)
- {
- t_float centroid = 0;
- int not = 0;
- for (j = 0, k = newphase; j < x->x_vibbins; j++)
- {
- centroid += phist->h_pitches[k];
- k--;
- if (k < 0) k = HISTORY-1;
- }
- centroid /= x->x_vibbins;
- for (j = 0, k = newphase; j < x->x_vibbins; j++)
- {
- /* calculate deviation from norm */
- t_float dev = centroid - phist->h_pitches[k];
- k--;
- if (k < 0) k = HISTORY-1;
- if (dev > x->x_vibdepth ||
- -dev > x->x_vibdepth) not = 1;
- }
- if (!not)
- {
- phist->h_pitch = phist->h_noted = centroid;
- }
- }
- }
- }
- return;
-
-nopow:
- for (i = 0; i < x->x_npitch; i++)
- {
- x->x_hist[i].h_pitch = x->x_hist[i].h_noted =
- x->x_hist[i].h_pitches[newphase] =
- x->x_hist[i].h_amps[newphase] = 0;
- x->x_hist[i].h_age = 0;
- }
- x->x_peaked = 1;
- x->x_dbage = 0;
-}
-
-void sigfiddle_debug(t_sigfiddle *x)
-{
- x->x_nprint = 1;
-}
-
-void sigfiddle_print(t_sigfiddle *x)
-{
- post("npoints %d,", 2 * x->x_hop);
- post("amp-range %f %f,", x->x_amplo, x->x_amphi);
- post("reattack %d %f,", x->x_attacktime, x->x_attackthresh);
- post("vibrato %d %f", x->x_vibtime, x->x_vibdepth);
- post("npartial %f", x->x_npartial);
- post("auto %d", x->x_auto);
-}
-
-void sigfiddle_amprange(t_sigfiddle *x, t_floatarg amplo, t_floatarg amphi)
-{
- if (amplo < 0) amplo = 0;
- if (amphi < amplo) amphi = amplo + 1;
- x->x_amplo = amplo;
- x->x_amphi = amphi;
-}
-
-void sigfiddle_reattack(t_sigfiddle *x,
- t_floatarg attacktime, t_floatarg attackthresh)
-{
- if (attacktime < 0) attacktime = 0;
- if (attackthresh <= 0) attackthresh = 1000;
- x->x_attacktime = attacktime;
- x->x_attackthresh = attackthresh;
- x->x_attackbins = (x->x_sr * 0.001 * attacktime) / x->x_hop;
- if (x->x_attackbins >= HISTORY) x->x_attackbins = HISTORY - 1;
-}
-
-void sigfiddle_vibrato(t_sigfiddle *x, t_floatarg vibtime, t_floatarg vibdepth)
-{
- if (vibtime < 0) vibtime = 0;
- if (vibdepth <= 0) vibdepth = 1000;
- x->x_vibtime = vibtime;
- x->x_vibdepth = vibdepth;
- x->x_vibbins = (x->x_sr * 0.001 * vibtime) / x->x_hop;
- if (x->x_vibbins >= HISTORY) x->x_vibbins = HISTORY - 1;
- if (x->x_vibbins < 1) x->x_vibbins = 1;
-}
-
-void sigfiddle_npartial(t_sigfiddle *x, t_floatarg npartial)
-{
- if (npartial < 0.1) npartial = 0.1;
- x->x_npartial = npartial;
-}
-
-void sigfiddle_auto(t_sigfiddle *x, t_floatarg f)
-{
- x->x_auto = (f != 0);
-}
-
-static void sigfiddle_freebird(t_sigfiddle *x)
-{
- if (x->x_inbuf)
- {
- freebytes(x->x_inbuf, sizeof(t_float) * x->x_hop);
- x->x_inbuf = 0;
- }
- if (x->x_lastanalysis)
- {
- freebytes(x->x_lastanalysis,
- sizeof(t_float) * (2 * x->x_hop + 4 * FILTSIZE));
- x->x_lastanalysis = 0;
- }
- if (x->x_spiral)
- {
- freebytes(x->x_spiral, sizeof(t_float) * 2 * x->x_hop);
- x->x_spiral = 0;
- }
- x->x_hop = 0;
-}
-
-int sigfiddle_setnpoints(t_sigfiddle *x, t_floatarg fnpoints)
-{
- int i, npoints = fnpoints;
- sigfiddle_freebird(x);
- if (npoints < MINPOINTS || npoints > MAXPOINTS)
- {
- error("fiddle~: npoints out of range; using %d",
- npoints = DEFAULTPOINTS);
- }
- if (npoints != (1 << sigfiddle_ilog2(npoints)))
- {
- error("fiddle~: npoints not a power of 2; using %d",
- npoints = (1 << sigfiddle_ilog2(npoints)));
- }
- x->x_hop = npoints >> 1;
- if (!(x->x_inbuf = (t_float *)getbytes(sizeof(t_float) * x->x_hop)))
- goto fail;
- if (!(x->x_lastanalysis = (t_float *)getbytes(
- sizeof(t_float) * (2 * x->x_hop + 4 * FILTSIZE))))
- goto fail;
- if (!(x->x_spiral = (t_float *)getbytes(sizeof(t_float) * 2 * x->x_hop)))
- goto fail;
- for (i = 0; i < x->x_hop; i++)
- x->x_inbuf[i] = 0;
- for (i = 0; i < npoints + 4 * FILTSIZE; i++)
- x->x_lastanalysis[i] = 0;
- for (i = 0; i < x->x_hop; i++)
- x->x_spiral[2*i] = cos((3.14159*i)/(npoints)),
- x->x_spiral[2*i+1] = -sin((3.14159*i)/(npoints));
- x->x_phase = 0;
- return (1);
-fail:
- sigfiddle_freebird(x);
- return (0);
-}
-
-int sigfiddle_doinit(t_sigfiddle *x, long npoints, long npitch,
- long npeakanal, long npeakout)
-{
- t_float *buf1, *buf2, *buf3;
- t_peakout *buf4;
- int i;
-
- if (!npeakanal && !npeakout) npeakanal = DEFNPEAK, npeakout = 0;
- if (!npeakanal < 0) npeakanal = 0;
- else if (npeakanal > MAXPEAK) npeakanal = MAXPEAK;
- if (!npeakout < 0) npeakout = 0;
- else if (npeakout > MAXPEAK) npeakout = MAXPEAK;
- if (npitch <= 0) npitch = 0;
- else if (npitch > MAXNPITCH) npitch = MAXNPITCH;
- if (npeakanal && !npitch) npitch = 1;
- if (!npoints)
- npoints = DEFAULTPOINTS;
- if (!sigfiddle_setnpoints(x, npoints))
- {
- error("fiddle~: out of memory");
- return (0);
- }
- if (!(buf4 = (t_peakout *)getbytes(sizeof(*buf4) * npeakout)))
- {
- sigfiddle_freebird(x);
- error("fiddle~: out of memory");
- return (0);
- }
- for (i = 0; i < npeakout; i++)
- buf4[i].po_freq = buf4[i].po_amp = 0;
- x->x_peakbuf = buf4;
-
- x->x_npeakout = npeakout;
- x->x_npeakanal = npeakanal;
- x->x_phase = 0;
- x->x_histphase = 0;
- x->x_sr = 44100; /* this and the next are filled in later */
- for (i = 0; i < MAXNPITCH; i++)
- {
- int j;
- x->x_hist[i].h_pitch = x->x_hist[i].h_noted = 0;
- x->x_hist[i].h_age = 0;
- x->x_hist[i].h_wherefrom = 0;
- x->x_hist[i].h_outlet = 0;
- for (j = 0; j < HISTORY; j++)
- x->x_hist[i].h_amps[j] = x->x_hist[i].h_pitches[j] = 0;
- }
- x->x_nprint = 0;
- x->x_npitch = npitch;
- for (i = 0; i < HISTORY; i++) x->x_dbs[i] = 0;
- x->x_dbage = 0;
- x->x_peaked = 0;
- x->x_auto = 1;
- x->x_amplo = DEFAMPLO;
- x->x_amphi = DEFAMPHI;
- x->x_attacktime = DEFATTACKTIME;
- x->x_attackbins = 1; /* real value calculated afterward */
- x->x_attackthresh = DEFATTACKTHRESH;
- x->x_vibtime = DEFVIBTIME;
- x->x_vibbins = 1; /* real value calculated afterward */
- x->x_vibdepth = DEFVIBDEPTH;
- x->x_npartial = 7;
- x->x_attackvalue = 0;
- return (1);
-}
-
- /* formalities for JMAX */
-
-#ifdef JMAX
-
-void sigfiddle_debug13(fts_object_t *o, int winlet, fts_symbol_t s, int ac, const fts_atom_t *at)
-{
- t_sigfiddle *x = (t_sigfiddle *)o;
- sigfiddle_debug(x);
-}
-
-void sigfiddle_print13(fts_object_t *o, int winlet, fts_symbol_t s,
- int ac, const fts_atom_t *at)
-{
- t_sigfiddle *x = (t_sigfiddle *)o;
- sigfiddle_print(x);
-}
-
-void sigfiddle_amprange13(fts_object_t *o, int winlet, fts_symbol_t s,
- int ac, const fts_atom_t *at)
-{
- t_sigfiddle *x = (t_sigfiddle *)o;
- t_float lo = (t_float) fts_get_float_arg(ac, at, 0, 0);
- t_float hi = (t_float) fts_get_float_arg(ac, at, 1, 0);
- sigfiddle_amprange(x, lo, hi);
-}
-
-void sigfiddle_reattack13(fts_object_t *o, int winlet, fts_symbol_t s,
- int ac, const fts_atom_t *at)
-{
- t_sigfiddle *x = (t_sigfiddle *)o;
- long msec = fts_get_float_arg(ac, at, 0, 0);
- t_float db = (t_float) fts_get_float_arg(ac, at, 1, 0);
- sigfiddle_reattack(x, msec, db);
-}
-
-void sigfiddle_vibrato13(fts_object_t *o, int winlet, fts_symbol_t s,
- int ac, const fts_atom_t *at)
-{
- t_sigfiddle *x = (t_sigfiddle *)o;
- long msec = fts_get_float_arg(ac, at, 0, 0);
- t_float halftones = (t_float) fts_get_float_arg(ac, at, 1, 0);
- sigfiddle_vibrato(x, msec, halftones);
-}
-
-void sigfiddle_npartial13(fts_object_t *o, int winlet, fts_symbol_t s,
- int ac, const fts_atom_t *at)
-{
- t_sigfiddle *x = (t_sigfiddle *)o;
- t_float npartial = (t_float) fts_get_float_arg(ac, at, 0, 0);
- sigfiddle_npartial(x, npartial);
-}
-
-
-void ftl_sigfiddle(fts_word_t *a)
-{
- t_sigfiddle *x = (t_sigfiddle *)fts_word_get_long(a);
- t_float *in = (t_float *)fts_word_get_long(a + 1);
- long n_tick = fts_word_get_long(a + 2);
-
- int count;
- t_float *fp, *fp2;
- for (count = 0, fp = x->x_inbuf + x->x_phase;
- count < n_tick; count++) *fp++ = *in++;
- if (fp == x->x_inbuf + x->x_hop)
- {
- sigfiddle_doit(x);
- x->x_phase = 0;
- fts_alarm_set_delay(&x->x_clock, 0L); /* output bang */
- fts_alarm_arm(&x->x_clock);
-
- if (x->x_nprint) x->x_nprint--;
- }
- else x->x_phase += n_tick;
-}
-
-void sigfiddle_put(fts_object_t *o, int winlet, fts_symbol_t *s, int ac, const fts_atom_t *at)
-{
- t_sigfiddle *x = (t_sigfiddle *)o;
- fts_dsp_descr_t *dsp = (fts_dsp_descr_t *)fts_get_long_arg(ac, at, 0, 0);
- fts_atom_t a[3];
-
- x->x_sr = fts_dsp_get_input_srate(dsp, 0);
- sigfiddle_reattack(x, x->x_attacktime, x->x_attackthresh);
- sigfiddle_vibrato(x, x->x_vibtime, x->x_vibdepth);
-
- fts_set_long(a, (long)x);
- fts_set_symbol(a+1, fts_dsp_get_input_name(dsp, 0));
- fts_set_long(a+2, fts_dsp_get_input_size(dsp, 0));
- dsp_add_funcall(dsp_symbol, 3, a);
-}
-
-void sigfiddle_tick(fts_alarm_t *alarm, void *p)
-{
- fts_object_t *o = (fts_object_t *)p;
- t_sigfiddle *x = (t_sigfiddle *)p;
-
- int i;
- t_pitchhist *ph;
- fts_outlet_float(o, OUTLETpower, x->x_dbs[x->x_histphase]);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- {
- fts_atom_t at[2];
- fts_set_float(at, ph->h_pitches[x->x_histphase]);
- fts_set_float(at+1, ph->h_amps[x->x_histphase]);
- fts_outlet_list(o, OUTLETmicropitch3 - i, 2, at);
- }
- if (x->x_attackvalue) fts_outlet_bang(o, OUTLETattack);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- if (ph->h_pitch) fts_outlet_float(o, OUTLETpitch, ph->h_pitch);
-}
-
-static void sigfiddle_delete(fts_object_t *o, int winlet, fts_symbol_t *s, int ac,
- const fts_atom_t *at)
-{
- t_sigfiddle *x = (t_sigfiddle *)o;
-
- fts_free(x->x_inbuf);
- fts_free(x->x_lastanalysis);
- fts_free(x->x_spiral);
- dsp_list_remove(o);
-}
-
-static void sigfiddle_init(fts_object_t *o, int winlet, fts_symbol_t *s, int ac, const fts_atom_t *at)
-{
- t_sigfiddle *x = (t_sigfiddle *)o;
- t_float *buf1, *buf2, *buf3;
- int i, hop;
- long npoints = fts_get_long_arg(ac, at, 1, 0);
- long npitch = fts_get_long_arg(ac, at, 2, 0);
- long npeakanal = fts_get_long_arg(ac, at, 3, 0);
- long npeakout = fts_get_long_arg(ac, at, 4, 0);
-
- if (!sigfiddle_doinit(x, npoints, npitch, npeakanal, npeakout))
- {
- post("fiddle~: initialization failed");
- return;
- }
- hop = npoints>>1;
- if (fts_fft_declaresize(hop) != fts_Success)
- post("fiddle~: bad FFT size");
-
- fts_alarm_init(&(x->x_clock), 0, sigfiddle_tick, x);
- dsp_list_insert(o);
-}
-
-static fts_status_t sigfiddle_instantiate(fts_class_t *cl, int ac,
- const fts_atom_t *at)
-{
- int i;
- fts_type_t a[5];
-
- fts_class_init(cl, sizeof(t_sigfiddle), 1, 6, 0); /* 1 inlet + 6 outlets */
-
- /* the system methods */
-
- a[0] = fts_Symbol;
- a[1] = fts_Long | fts_OptArg;
- a[2] = fts_Long | fts_OptArg;
- fts_method_define(cl, fts_SystemInlet, fts_s_init, sigfiddle_init, 3, a);
-
- fts_method_define(cl, fts_SystemInlet, fts_s_delete, sigfiddle_delete, 0, a);
- a[0] = fts_Object;
- fts_method_define(cl, fts_SystemInlet, fts_s_put, sigfiddle_put, 1, a);
-
- /* class' own methods */
- fts_method_define(cl, 0, fts_new_symbol("print"), sigfiddle_print13, 0, a);
- fts_method_define(cl, 0, fts_new_symbol("debug"), sigfiddle_debug13, 0, a);
- fts_method_define(cl, 0, fts_new_symbol("amp-range"), sigfiddle_amprange13,
- 0, a);
- fts_method_define(cl, 0, fts_new_symbol("reattack"), sigfiddle_reattack13,
- 0, a);
- fts_method_define(cl, 0, fts_new_symbol("vibrato"), sigfiddle_vibrato13,
- 0, a);
- fts_method_define(cl, 0, fts_new_symbol("npartial"), sigfiddle_npartial13,
- 0, a);
-
- /* classes signal inlets */
- dsp_sig_inlet(cl, 0); /* declare signal input #0 */
-
- /* classes outlets */
- a[0] = fts_Float;
- fts_outlet_type_define(cl, OUTLETpitch, fts_s_float, 1, a); /* declare outlet #0 */
- fts_outlet_type_define(cl, OUTLETattack, fts_s_bang, 0, a); /* declare outlet #1 */
- a[0] = fts_VarArgs;
- fts_outlet_type_define(cl, OUTLETmicropitch1, fts_s_list, 1, a); /* declare outlet #2 */
- fts_outlet_type_define(cl, OUTLETmicropitch2, fts_s_list, 1, a); /* declare outlet #3 */
- fts_outlet_type_define(cl, OUTLETmicropitch3, fts_s_list, 1, a); /* declare outlet #4 */
- a[0] = fts_Float;
- fts_outlet_type_define(cl, OUTLETpower, fts_s_float, 1, a); /* declare outlet #5 */
-
- dsp_symbol = fts_new_symbol("fiddle");
- dsp_declare_function(dsp_symbol, ftl_sigfiddle);
-
- /* DSP properties */
-
- fts_class_put_prop(cl, fts_s_dsp_is_sink, fts_true);
-
- return(fts_Success);
-}
-
-void fiddle_config(void)
-{
- sys_log(fiddle_version);
- fts_metaclass_create(fts_new_symbol(CLASSNAME), sigfiddle_instantiate, fts_always_equiv);
-}
-
-fts_module_t fiddle_module =
- {"fiddle", "sonic meat fiddle", fiddle_config, 0};
-
-#endif /* JMAX */
-
-#ifdef PD
-
-static t_int *fiddle_perform(t_int *w)
-{
- t_float *in = (t_float *)(w[1]);
- t_sigfiddle *x = (t_sigfiddle *)(w[2]);
- int n = (int)(w[3]);
- int count;
- t_float *fp;
- if (!x->x_hop)
- goto nono;
- for (count = 0, fp = x->x_inbuf + x->x_phase; count < n; count++)
- *fp++ = *in++;
- if (fp == x->x_inbuf + x->x_hop)
- {
- sigfiddle_doit(x);
- x->x_phase = 0;
- if (x->x_auto) clock_delay(x->x_clock, 0L);
- if (x->x_nprint) x->x_nprint--;
- }
- else x->x_phase += n;
-nono:
- return (w+4);
-}
-
-void sigfiddle_dsp(t_sigfiddle *x, t_signal **sp)
-{
- x->x_sr = sp[0]->s_sr;
- sigfiddle_reattack(x, x->x_attacktime, x->x_attackthresh);
- sigfiddle_vibrato(x, x->x_vibtime, x->x_vibdepth);
- dsp_add(fiddle_perform, 3, sp[0]->s_vec, x, sp[0]->s_n);
-}
-
- /* This is the callback function for the clock, but also acts as
- the "bang" method; you can leave "auto" on to get this called
- automatically (the default) or turn auto off and bang it yourself. */
-
-void sigfiddle_bang(t_sigfiddle *x)
-{
- int i;
- t_pitchhist *ph;
- if (x->x_npeakout)
- {
- int npeakout = x->x_npeakout;
- t_peakout *po;
- for (i = 0, po = x->x_peakbuf; i < npeakout; i++, po++)
- {
- t_atom at[3];
- SETFLOAT(at, i+1);
- SETFLOAT(at+1, po->po_freq);
- SETFLOAT(at+2, po->po_amp);
- outlet_list(x->x_peakout, 0, 3, at);
- }
- }
- outlet_float(x->x_envout, x->x_dbs[x->x_histphase]);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- {
- t_atom at[2];
- SETFLOAT(at, ph->h_pitches[x->x_histphase]);
- SETFLOAT(at+1, ph->h_amps[x->x_histphase]);
- outlet_list(ph->h_outlet, 0, 2, at);
- }
- if (x->x_attackvalue) outlet_bang(x->x_attackout);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- if (ph->h_pitch) outlet_float(x->x_noteout, ph->h_pitch);
-}
-
-void sigfiddle_ff(t_sigfiddle *x) /* cleanup on free */
-{
- if (x->x_inbuf)
- {
- freebytes(x->x_inbuf, sizeof(t_float) * x->x_hop);
- freebytes(x->x_lastanalysis, sizeof(t_float) * (2*x->x_hop + 4 * FILTSIZE));
- freebytes(x->x_spiral, sizeof(t_float) * 2*x->x_hop);
- freebytes(x->x_peakbuf, sizeof(*x->x_peakbuf) * x->x_npeakout);
- clock_free(x->x_clock);
- }
-}
-
-static t_class *sigfiddle_class;
-
-void *sigfiddle_new(t_floatarg npoints, t_floatarg npitch,
- t_floatarg fnpeakanal, t_floatarg fnpeakout)
-{
- t_sigfiddle *x = (t_sigfiddle *)pd_new(sigfiddle_class);
- int i;
- int npeakanal = fnpeakanal, npeakout = fnpeakout;
-
-
- if (!sigfiddle_doinit(x, npoints, npitch,
- npeakanal, npeakout))
- {
- x->x_inbuf = 0; /* prevent the free routine from cleaning up */
- pd_free(&x->x_ob.ob_pd);
- return (0);
- }
- x->x_noteout = outlet_new(&x->x_ob, gensym("float"));
- x->x_attackout = outlet_new(&x->x_ob, gensym("bang"));
- for (i = 0; i < x->x_npitch; i++)
- x->x_hist[i].h_outlet = outlet_new(&x->x_ob, gensym("list"));
- x->x_envout = outlet_new(&x->x_ob, gensym("float"));
- if (x->x_npeakout)
- x->x_peakout = outlet_new(&x->x_ob, gensym("list"));
- else x->x_peakout = 0;
- x->x_clock = clock_new(&x->x_ob.ob_pd, (t_method)sigfiddle_bang);
- return (x);
-}
-
-void fiddle_tilde_setup(void)
-{
- sigfiddle_class = class_new(gensym("fiddle~"), (t_newmethod)sigfiddle_new,
- (t_method)sigfiddle_ff, sizeof(t_sigfiddle), 0,
- A_DEFFLOAT, A_DEFFLOAT, A_DEFFLOAT, A_DEFFLOAT, 0);
- class_addmethod(sigfiddle_class, (t_method)sigfiddle_dsp,
- gensym("dsp"), 0);
- class_addmethod(sigfiddle_class, (t_method)sigfiddle_debug,
- gensym("debug"), 0);
- class_addmethod(sigfiddle_class, (t_method)sigfiddle_setnpoints,
- gensym("npoints"), A_FLOAT, 0);
- class_addmethod(sigfiddle_class, (t_method)sigfiddle_amprange,
- gensym("amp-range"), A_FLOAT, A_FLOAT, 0);
- class_addmethod(sigfiddle_class, (t_method)sigfiddle_reattack,
- gensym("reattack"), A_FLOAT, A_FLOAT, 0);
- class_addmethod(sigfiddle_class, (t_method)sigfiddle_vibrato,
- gensym("vibrato"), A_FLOAT, A_FLOAT, 0);
- class_addmethod(sigfiddle_class, (t_method)sigfiddle_npartial,
- gensym("npartial"), A_FLOAT, 0);
- class_addmethod(sigfiddle_class, (t_method)sigfiddle_auto,
- gensym("auto"), A_FLOAT, 0);
- class_addmethod(sigfiddle_class, (t_method)sigfiddle_print,
- gensym("print"), 0);
- class_addmethod(sigfiddle_class, nullfn, gensym("signal"), 0);
- class_addbang(sigfiddle_class, sigfiddle_bang);
- class_addcreator((t_newmethod)sigfiddle_new, gensym("fiddle"),
- A_DEFFLOAT, A_DEFFLOAT, A_DEFFLOAT, A_DEFFLOAT, 0);
- post(fiddle_version);
-}
-
-void fiddle_setup(void)
-{
- fiddle_tilde_setup();
-}
-#endif /* PD */
-
-#ifdef MAX26
-
-void cu_fiddle(t_float *in1, t_sigfiddle *x, int n)
-{
- int count;
- t_float *fp, *fp2;
- for (count = 0, fp = x->x_inbuf + x->x_phase;
- count < n; count++) *fp++ = *in1++;
- if (fp == x->x_inbuf + x->x_hop)
- {
- sigfiddle_doit(x);
- x->x_phase = 0;
- if (x->x_auto) clock_delay(x->x_clock, 0L);
- if (x->x_nprint) x->x_nprint--;
- }
- else x->x_phase += n;
-}
-
-void sigfiddle_put(t_sigfiddle *x, long whether)
-{
- if (whether)
- {
- u_stdout(x);
- x->x_sr = x->x_io[0]->s_sr;
- sigfiddle_reattack(x, x->x_attacktime, x->x_attackthresh);
- sigfiddle_vibrato(x, x->x_vibtime, x->x_vibdepth);
- dspchain_addc(cu_fiddle, 3,
- x->x_io[0]->s_shit, x, x->x_io[0]->s_n);
- }
-}
-
-void sigfiddle_tick(t_sigfiddle *x) /* callback function for the clock */
-{
- int i;
- t_pitchhist *ph;
- outlet_float(x->x_envout, x->x_dbs[x->x_histphase]);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- {
- t_atom at[2];
- SETFLOAT(at, ph->h_pitches[x->x_histphase]);
- SETFLOAT(at+1, ph->h_amps[x->x_histphase]);
- outlet_list(ph->h_outlet, NIL, 2, at);
- }
- if (x->x_attackvalue) outlet_bang(x->x_attackout);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- if (ph->h_pitch) outlet_float(x->x_noteout, ph->h_pitch);
-}
-
-void sigfiddle_ff(t_sigfiddle *x) /* cleanup on free */
-{
- if (x->x_inbuf)
- {
- freebytes(x->x_inbuf, sizeof(t_float) * x->x_hop);
- freebytes(x->x_lastanalysis, sizeof(t_float) * (2*x->x_hop + 4 * FILTSIZE));
- freebytes(x->x_spiral, sizeof(t_float) * 2*x->x_hop);
- clock_free(x->x_clock);
- u_clean(x);
- }
-}
-
-t_externclass *sigfiddle_class;
-
-void *sigfiddle_new(long npoints, long npitch,
- long npeakanal, long npeakout)
-{
- t_sigfiddle *x = (t_sigfiddle *)obj_new(&sigfiddle_class, 0);
- int i;
-
- if (!sigfiddle_doinit(x, npoints, npitch, npeakanal, npeakout))
- {
- x->x_inbuf = 0; /* prevent the free routine from cleaning up */
- obj_free(x);
- return (0);
- }
- u_setup(x, IN1, OUT0);
- x->x_envout = outlet_new(x, gensym("float"));
- for (i = 0; i < x->x_npitch; i++)
- x->x_hist[i].h_outlet = outlet_new(x, gensym("list"));
- x->x_attackout = outlet_new(x, gensym("bang"));
- x->x_noteout = outlet_new(x, gensym("float"));
- x->x_clock = clock_new(x, sigfiddle_tick);
- return (x);
-}
-
-void fiddle_setup()
-{
- c_extern(&sigfiddle_class, sigfiddle_new, sigfiddle_ff,
- gensym("fiddle"), sizeof(t_sigfiddle), 0, A_DEFLONG, A_DEFLONG,
- A_DEFLONG, A_DEFLONG, 0);
- c_addmess(sigfiddle_put, gensym("put"), A_CANT, 0);
- c_addmess(sigfiddle_debug, gensym("debug"), 0);
- c_addmess(sigfiddle_amprange, gensym("amp-range"), A_FLOAT, A_FLOAT, 0);
- c_addmess(sigfiddle_reattack, gensym("reattack"), A_FLOAT, A_FLOAT, 0);
- c_addmess(sigfiddle_vibrato, gensym("vibrato"), A_LONG, A_FLOAT, 0);
- c_addmess(sigfiddle_npartial, gensym("npartial"), A_FLOAT, 0);
- c_addmess(sigfiddle_print, gensym("print"), 0);
- u_inletmethod(0); /* one signal input */
-#ifdef MAX
- post(fiddle_version);
-#endif
-}
-
-#endif /* MAX26 */
-
-/************* Beginning of MSP Code ******************************/
-
-#ifdef MSP
-
-static t_int *fiddle_perform(t_int *w)
-{
- t_float *in = (t_float *)(w[1]);
- t_sigfiddle *x = (t_sigfiddle *)(w[2]);
- int n = (int)(w[3]);
- int count,inc = x->x_downsample;
- t_float *fp;
-
- if (x->x_obj.z_disabled)
- goto skip;
- for (count = 0, fp = x->x_inbuf + x->x_phase; count < n; count+=inc) {
- *fp++ = *in;
- in += inc;
- }
- if (fp == x->x_inbuf + x->x_hop)
- {
- sigfiddle_doit(x);
- x->x_phase = 0;
- if (x->x_auto) clock_delay(x->x_clock, 0L);
- if (x->x_nprint) x->x_nprint--;
- }
- else x->x_phase += n;
-skip:
- return (w+4);
-}
-
-void sigfiddle_dsp(t_sigfiddle *x, t_signal **sp)
-{
- if (sp[0]->s_n > x->x_hop) {
- x->x_downsample = sp[0]->s_n / x->x_hop;
- post("* warning: fiddle~: will downsample input by %ld",x->x_downsample);
- x->x_sr = sp[0]->s_sr / x->x_downsample;
- } else {
- x->x_downsample = 1;
- x->x_sr = sp[0]->s_sr;
- }
- sigfiddle_reattack(x, x->x_attacktime, x->x_attackthresh);
- sigfiddle_vibrato(x, x->x_vibtime, x->x_vibdepth);
- dsp_add(fiddle_perform, 3, sp[0]->s_vec, x, sp[0]->s_n);
-}
-
-void sigfiddle_tick(t_sigfiddle *x) /* callback function for the clock MSP*/
-{
- int i;
- t_pitchhist *ph;
- if (x->x_npeakout)
- {
- int npeakout = x->x_npeakout;
- t_peakout *po;
- for (i = 0, po = x->x_peakbuf; i < npeakout; i++, po++)
- {
- t_atom at[3];
- SETINT(at, i+1);
- SETFLOAT(at+1, po->po_freq);
- SETFLOAT(at+2, po->po_amp);
- outlet_list(x->x_peakout, 0, 3, at);
- }
- }
- outlet_float(x->x_envout, x->x_dbs[x->x_histphase]);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- {
- t_atom at[2];
- SETFLOAT(at, ph->h_pitches[x->x_histphase]);
- SETFLOAT(at+1, ph->h_amps[x->x_histphase]);
- outlet_list(ph->h_outlet, 0, 2, at);
- }
- if (x->x_attackvalue) outlet_bang(x->x_attackout);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- if (ph->h_pitch) outlet_float(x->x_noteout, ph->h_pitch);
-}
-
-void sigfiddle_bang(t_sigfiddle *x)
-{
- int i;
- t_pitchhist *ph;
- if (x->x_npeakout)
- {
- int npeakout = x->x_npeakout;
- t_peakout *po;
- for (i = 0, po = x->x_peakbuf; i < npeakout; i++, po++)
- {
- t_atom at[3];
- SETLONG(at, i+1);
- SETFLOAT(at+1, po->po_freq);
- SETFLOAT(at+2, po->po_amp);
- outlet_list(x->x_peakout, 0, 3, at);
- }
- }
- outlet_float(x->x_envout, x->x_dbs[x->x_histphase]);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- {
- t_atom at[2];
- SETFLOAT(at, ph->h_pitches[x->x_histphase]);
- SETFLOAT(at+1, ph->h_amps[x->x_histphase]);
- outlet_list(ph->h_outlet, 0, 2, at);
- }
- if (x->x_attackvalue) outlet_bang(x->x_attackout);
- for (i = 0, ph = x->x_hist; i < x->x_npitch; i++, ph++)
- if (ph->h_pitch) outlet_float(x->x_noteout, ph->h_pitch);
-}
-
-
-void sigfiddle_ff(t_sigfiddle *x) /* cleanup on free MSP */
-{
-
- if (x->x_inbuf)
- {
- t_freebytes(x->x_inbuf, sizeof(t_float) * x->x_hop);
- t_freebytes(x->x_lastanalysis, sizeof(t_float) * (2*x->x_hop + 4 *
-FILTSIZE));
- t_freebytes(x->x_spiral, sizeof(t_float) * 2*x->x_hop);
- t_freebytes(x->x_peakbuf, sizeof(*x->x_peakbuf) * x->x_npeakout);
- }
- dsp_free((t_pxobject *)x);
-}
-
-void *sigfiddle_class;
-
-void *sigfiddle_new(long npoints, long npitch,
- long npeakanal, long npeakout)
-{
- t_sigfiddle *x = (t_sigfiddle *)newobject(sigfiddle_class);
- int i;
-
- if (!sigfiddle_doinit(x, npoints, npitch, npeakanal, npeakout))
- {
- x->x_inbuf = 0; /* prevent the free routine from cleaning up */
- return (0);
- }
- dsp_setup((t_pxobject *)x,1);
-
- x->x_clock = clock_new(x, (method)sigfiddle_tick);
- if (x->x_npeakout)
- x->x_peakout = listout((t_object *)x);
- else x->x_peakout = 0;
- x->x_envout = floatout((t_object *)x);
- for (i = 0; i < x->x_npitch; i++)
- x->x_hist[i].h_outlet = listout((t_object *)x);
- x->x_attackout = bangout((t_object *)x);
- x->x_noteout = floatout((t_object *)x);
- return (x);
-
-
-}
-
-void main()
-{
- setup(&sigfiddle_class, sigfiddle_new, (method)sigfiddle_ff,
- (short)sizeof(t_sigfiddle), 0L, A_DEFLONG, A_DEFLONG,
-A_DEFLONG, A_DEFLONG, 0);
- addmess((method)sigfiddle_dsp, "dsp",
- A_CANT, 0);
- addmess((method)sigfiddle_debug, "debug", 0);
- addmess((method)sigfiddle_setnpoints, "npoints", A_FLOAT, 0);
- addmess((method)sigfiddle_amprange, "amp-range", A_FLOAT, A_FLOAT, 0);
- addmess((method)sigfiddle_reattack, "reattack", A_FLOAT, A_FLOAT, 0);
- addmess((method)sigfiddle_vibrato, "vibrato", A_FLOAT,
-A_FLOAT, 0);
- addmess((method)sigfiddle_npartial, "npartial", A_FLOAT, 0);
- addmess((method)sigfiddle_auto, "auto",
- A_FLOAT, 0);
- addmess((method)sigfiddle_print, "print", 0);
- addmess((method)sigfiddle_assist, "assist",
- A_CANT, 0);
- addbang((method)sigfiddle_bang);
- dsp_initclass();
- rescopy('STR#',3748);
- post(fiddle_version);
-}
-
-void sigfiddle_assist(t_sigfiddle *x, void *b, long m, long a, char *s)
-{
- assist_string(3748,m,a,1,2,s);
-}
-
-void msp_fft(t_float *buf, long np, long inv)
-{
- t_float *src,*real,*rp,*imag,*ip;
- long i;
-
- /*
- // because this fft algorithm uses separate real and imaginary
- // buffers
- // we must split the real and imaginary parts into two buffers,
- // then do the opposite on output
- // a more ambitious person would either do an in-place conversion
- // or rewrite the fft algorithm
- */
-
- real = rp = msp_ffttemp;
- imag = ip = real + MAXPOINTS;
- src = buf;
- for (i = 0; i < np; i++) {
- *rp++ = *src++;
- *ip++ = *src++;
- }
- if (inv)
- ifft(np,real,imag);
- else
- fft(np,real,imag);
- rp = real;
- ip = imag;
- src = buf;
- for (i = 0; i < np; i++) {
- *src++ = *rp++;
- *src++ = *ip++;
- }
-}
-
-#endif /* MSP */