From f9e3fe480e7e3c448a626d508f1976f4fec2f759 Mon Sep 17 00:00:00 2001 From: Hans-Christoph Steiner Date: Tue, 9 Feb 2010 21:05:37 +0000 Subject: convert file layout to libdir style and setup everything with the new template Makefile+debian stuff svn path=/trunk/externals/sigpack/; revision=13153 --- freqshift~.c | 197 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 197 insertions(+) create mode 100644 freqshift~.c (limited to 'freqshift~.c') diff --git a/freqshift~.c b/freqshift~.c new file mode 100644 index 0000000..026e4cb --- /dev/null +++ b/freqshift~.c @@ -0,0 +1,197 @@ +/* sIgpAck + * for + * pure-data + * www.weiss-archiv.de */ + +#include "m_pd.h" +#include +#ifdef _MSC_VER +#pragma warning( disable : 4244 ) +#pragma warning( disable : 4305 ) +#define M_PI 3.14159265358979323846 +#endif + +/* ------------------------ freqshift~ ----------------------------- */ +/* frequency shifter */ +/* code from swh_plugins by steve harris www.plugins.org.uk */ + +#define SIN_T_SIZE 64 +#define D_SIZE 256 +#define NZEROS 200 + +static t_class *freqshift_tilde_class; + +typedef struct _freqshift_tilde +{ + t_object x_obj; + t_float x_shift;//[0 - 5000] + float *x_delay; + unsigned int x_dptr; + t_float x_fs; + t_float x_last_shift; + t_float x_phi; + float *x_sint; + float x_f; +} t_freqshift_tilde; + +static void *freqshift_tilde_new(t_floatarg shift) +{ + unsigned int i; + + t_freqshift_tilde *x = (t_freqshift_tilde *)pd_new(freqshift_tilde_class); + //x->x_shift = shift; + outlet_new(&x->x_obj, gensym("signal")); + outlet_new(&x->x_obj, gensym("signal")); + floatinlet_new(&x->x_obj, &x->x_shift); + x->x_fs = sys_getsr(); + x->x_delay = (float *)getbytes(D_SIZE * sizeof(float)); + x->x_sint = (float *)getbytes(SIN_T_SIZE * sizeof(float)); + x->x_dptr = 0; + x->x_phi = 0.0f; + x->x_last_shift = 0.0f; + x->x_f = 0; + for (i = 0; i < SIN_T_SIZE; i++) { + x->x_sint[i] = sin(2.0f * M_PI * (float)i / (float)SIN_T_SIZE); + } + if (shift) x->x_shift = shift; + else x->x_shift = 0; + return (x); +} + +/* The non-zero taps of the Hilbert transformer */ +static float xcoeffs[] = { + +0.0008103736f, +0.0008457886f, +0.0009017196f, +0.0009793364f, + +0.0010798341f, +0.0012044365f, +0.0013544008f, +0.0015310235f, + +0.0017356466f, +0.0019696659f, +0.0022345404f, +0.0025318040f, + +0.0028630784f, +0.0032300896f, +0.0036346867f, +0.0040788644f, + +0.0045647903f, +0.0050948365f, +0.0056716186f, +0.0062980419f, + +0.0069773575f, +0.0077132300f, +0.0085098208f, +0.0093718901f, + +0.0103049226f, +0.0113152847f, +0.0124104218f, +0.0135991079f, + +0.0148917649f, +0.0163008758f, +0.0178415242f, +0.0195321089f, + +0.0213953037f, +0.0234593652f, +0.0257599469f, +0.0283426636f, + +0.0312667947f, +0.0346107648f, +0.0384804823f, +0.0430224431f, + +0.0484451086f, +0.0550553725f, +0.0633242001f, +0.0740128560f, + +0.0884368322f, +0.1090816773f, +0.1412745301f, +0.1988673273f, + +0.3326528346f, +0.9997730178f, -0.9997730178f, -0.3326528346f, + -0.1988673273f, -0.1412745301f, -0.1090816773f, -0.0884368322f, + -0.0740128560f, -0.0633242001f, -0.0550553725f, -0.0484451086f, + -0.0430224431f, -0.0384804823f, -0.0346107648f, -0.0312667947f, + -0.0283426636f, -0.0257599469f, -0.0234593652f, -0.0213953037f, + -0.0195321089f, -0.0178415242f, -0.0163008758f, -0.0148917649f, + -0.0135991079f, -0.0124104218f, -0.0113152847f, -0.0103049226f, + -0.0093718901f, -0.0085098208f, -0.0077132300f, -0.0069773575f, + -0.0062980419f, -0.0056716186f, -0.0050948365f, -0.0045647903f, + -0.0040788644f, -0.0036346867f, -0.0032300896f, -0.0028630784f, + -0.0025318040f, -0.0022345404f, -0.0019696659f, -0.0017356466f, + -0.0015310235f, -0.0013544008f, -0.0012044365f, -0.0010798341f, + -0.0009793364f, -0.0009017196f, -0.0008457886f, -0.0008103736f, +}; + +static float f_clamp(float x, float a, float b) +{ + const float x1 = fabs(x - a); + const float x2 = fabs(x - b); + + x = x1 + a + b; + x -= x2; + x *= 0.5; + + return x; +} + +// Round float to int using IEEE int* hack +static int f_round(float f) { + f += (3<<22); + return *((int*)&f) - 0x4b400000; +} + +// Cubic interpolation function +static float cube_interp(const float fr, const float inm1, const float + in, const float inp1, const float inp2) +{ + return in + 0.5f * fr * (inp1 - inm1 + + fr * (4.0f * inp1 + 2.0f * inm1 - 5.0f * in - inp2 + + fr * (3.0f * (in - inp1) - inm1 + inp2))); +} + +static t_int *freqshift_tilde_perform(t_int *w) +{ + t_freqshift_tilde *x = (t_freqshift_tilde *)(w[1]); + t_float *in = (t_float *)(w[2]); + t_float *out1 = (t_float *)(w[3]); + t_float *out2 = (t_float *)(w[4]); + int n = (int)(w[5]); + float f, hilb, rm1, rm2, frac_p; + float shift_i = x->x_last_shift; + float sample_count = sys_getblksize(); + unsigned int i; + int int_p; + const float shift_c = f_clamp(x->x_shift, 0.0f, 10000.0f); + const float shift_inc = (shift_c - x->x_last_shift) / (float)sample_count; + const float freq_fix = (float)SIN_T_SIZE / x->x_fs; + while (n--) + { + f = *in++; + x->x_delay[x->x_dptr] = f; + /* Perform the Hilbert FIR convolution + * (probably FFT would be faster) */ + hilb = 0.0f; + for (i = 0; i <= NZEROS/2; i++) { + hilb += (xcoeffs[i] * x->x_delay[(x->x_dptr - i*2) & (D_SIZE - 1)]); + } + + /* Calcuate the table positions for the sine modulator */ + int_p = f_round(floor(x->x_phi)); + + /* Calculate ringmod1, the transformed input modulated with a shift Hz + * sinewave. This creates a +180 degree sideband at source-shift Hz and + * a 0 degree sindeband at source+shift Hz */ + frac_p = x->x_phi - int_p; + rm1 = hilb * cube_interp(frac_p, x->x_sint[int_p], x->x_sint[int_p+1], + x->x_sint[int_p+2], x->x_sint[int_p+3]); + + /* Calcuate the table positions for the cosine modulator */ + int_p = (int_p + SIN_T_SIZE / 4) & (SIN_T_SIZE - 1); + + /* Calculate ringmod2, the delayed input modulated with a shift Hz + * cosinewave. This creates a 0 degree sideband at source+shift Hz + * and a -180 degree sindeband at source-shift Hz */ + rm2 = x->x_delay[(x->x_dptr - 100) & (D_SIZE - 1)] * cube_interp(frac_p, + x->x_sint[int_p], x->x_sint[int_p+1], x->x_sint[int_p+2], x->x_sint[int_p+3]); + + /* Output the sum and differences of the ringmods. The +/-180 degree + * sidebands cancel (more of less) and just leave the shifted + * components */ + *out1++ = (rm2 - rm1) * 0.5f; /*downshifting*/ + *out2++ = (rm2 + rm1) * 0.5f; /*upshifting*/ + + x->x_dptr = (x->x_dptr + 1) & (D_SIZE - 1); + x->x_phi += shift_i * freq_fix; + while (x->x_phi > SIN_T_SIZE) { + x->x_phi -= SIN_T_SIZE; + } + shift_i += shift_inc; + } + return (w+6); +} + +static void freqshift_tilde_dsp(t_freqshift_tilde *x, t_signal **sp) +{ + dsp_add(freqshift_tilde_perform, 5, x, sp[0]->s_vec, sp[1]->s_vec, sp[2]->s_vec, sp[0]->s_n); +} + +static void freqshift_tilde_free(t_freqshift_tilde *x) +{ + if(x->x_delay) + freebytes(x->x_delay, D_SIZE * sizeof(float)); + if(x->x_sint) + freebytes(x->x_sint, SIN_T_SIZE + 4 * sizeof(float)); +} + +void freqshift_tilde_setup(void) +{ + freqshift_tilde_class = class_new(gensym("freqshift~"), (t_newmethod)freqshift_tilde_new, (t_method)freqshift_tilde_free, + sizeof(t_freqshift_tilde), 0, A_DEFFLOAT, 0); + CLASS_MAINSIGNALIN(freqshift_tilde_class, t_freqshift_tilde, x_f); + class_addmethod(freqshift_tilde_class, (t_method)freqshift_tilde_dsp, gensym("dsp"), 0); +} -- cgit v1.2.1