Autres/audacity
1
0
Fork 0
mirror of https://github.com/cookiengineer/audacity synced 2025-11-24 22:30:21 +01:00
Code Issues Releases Wiki Activity

Update Nyquist to v3.09.

Browse Source
This commit is contained in:
Leland Lucius
2015-04-07 22:10:17 -05:00
parent f88b27e6d8
commit 9fb0ce5b82
358 changed files with 26327 additions and 7043 deletions
Download Patch File Download Diff File Expand all files Collapse all files

361
lib-src/libnyquist/nyquist/tran/alpassvc.c
View File

@@ -9,11 +9,12 @@
#include "cext.h"
#include "alpassvc.h"
void alpassvc_free();
void alpassvc_free(snd_susp_type a_susp);
typedef struct alpassvc_susp_struct {
snd_susp_node susp;
boolean started;
long terminate_cnt;
sound_type input;
long input_cnt;
@@ -22,6 +23,15 @@ typedef struct alpassvc_susp_struct {
long delaysnd_cnt;
sample_block_values_type delaysnd_ptr;
/* support for interpolation of delaysnd */
sample_type delaysnd_x1_sample;
double delaysnd_pHaSe;
double delaysnd_pHaSe_iNcR;
/* support for ramp between samples of delaysnd */
double output_per_delaysnd;
long delaysnd_n;
float delay_scale_factor;
double feedback;
long buflen;
@@ -31,8 +41,9 @@ typedef struct alpassvc_susp_struct {
} alpassvc_susp_node, *alpassvc_susp_type;
void alpassvc_nn_fetch(register alpassvc_susp_type susp, snd_list_type snd_list)
void alpassvc_nn_fetch(snd_susp_type a_susp, snd_list_type snd_list)
{
alpassvc_susp_type susp = (alpassvc_susp_type) a_susp;
int cnt = 0; /* how many samples computed */
int togo;
int n;
@@ -69,6 +80,7 @@ void alpassvc_nn_fetch(register alpassvc_susp_type susp, snd_list_type snd_list)
if (susp->terminate_cnt != UNKNOWN &&
susp->terminate_cnt <= susp->susp.current + cnt + togo) {
togo = susp->terminate_cnt - (susp->susp.current + cnt);
if (togo < 0) togo = 0; /* avoids rounding errros */
if (togo == 0) break;
}
@@ -82,38 +94,38 @@ void alpassvc_nn_fetch(register alpassvc_susp_type susp, snd_list_type snd_list)
input_ptr_reg = susp->input_ptr;
out_ptr_reg = out_ptr;
if (n) do { /* the inner sample computation loop */
register sample_type y, z, delaysamp;
register int delayi;
register sample_type *yptr;
register sample_type y, z, delaysamp;
register int delayi;
register sample_type *yptr;
/* compute where to read y, we want y to be delay_snd samples
* after delay_ptr, where we write the new sample. First,
* conver from seconds to samples. Note: don't use actual sound_type
* names in comments! The translator isn't smart enough.
*/
delaysamp = *delaysnd_ptr_reg++ * delay_scale_factor_reg;
delayi = (int) delaysamp; /* get integer part */
delaysamp = delaysamp - delayi; /* get phase */
yptr = delayptr_reg + buflen_reg - (delayi + 1);
if (yptr >= endptr_reg) yptr -= buflen_reg;
/* now get y, the out-put of the delay, using interpolation */
/* note that as phase increases, we use more of yptr[0] because
positive phase means longer buffer means read earlier sample */
y = (float) ((yptr[0] * delaysamp) + (yptr[1] * (1.0 - delaysamp)));
/* WARNING: no check to keep delaysamp in range, so
do this in LISP */
/* compute where to read y, we want y to be delay_snd samples
* after delay_ptr, where we write the new sample. First,
* conver from seconds to samples. Note: don't use actual sound_type
* names in comments! The translator isn't smart enough.
*/
delaysamp = *delaysnd_ptr_reg++ * delay_scale_factor_reg;
delayi = (int) delaysamp; /* get integer part */
delaysamp = delaysamp - delayi; /* get phase */
yptr = delayptr_reg + buflen_reg - (delayi + 1);
if (yptr >= endptr_reg) yptr -= buflen_reg;
/* now get y, the out-put of the delay, using interpolation */
/* note that as phase increases, we use more of yptr[0] because
positive phase means longer buffer means read earlier sample */
y = (float) ((yptr[0] * delaysamp) + (yptr[1] * (1.0 - delaysamp)));
/* WARNING: no check to keep delaysamp in range, so do this in LISP */
*delayptr_reg++ = z = (sample_type) (feedback_reg * y + *input_ptr_reg++);
/* Time out to update the buffer:
* this is a tricky buffer: buffer[0] == buffer[bufflen]
* the logical length is bufflen, but the actual length
* is bufflen + 1 to allow for a repeated sample at the
* end. This allows for efficient interpolation.
*/
if (delayptr_reg > endptr_reg) {
delayptr_reg = susp->delaybuf;
*delayptr_reg++ = *endptr_reg;
}
*out_ptr_reg++ = (sample_type) (y - feedback_reg * z);;
*delayptr_reg++ = z = (sample_type) (feedback_reg * y + *input_ptr_reg++);
/* Time out to update the buffer:
* this is a tricky buffer: buffer[0] == buffer[bufflen]
* the logical length is bufflen, but the actual length
* is bufflen + 1 to allow for a repeated sample at the
* end. This allows for efficient interpolation.
*/
if (delayptr_reg > endptr_reg) {
delayptr_reg = susp->delaybuf;
*delayptr_reg++ = *endptr_reg;
}
*out_ptr_reg++ = (sample_type) (y - feedback_reg * z);
} while (--n); /* inner loop */
susp->buflen = buflen_reg;
@@ -138,9 +150,11 @@ void alpassvc_nn_fetch(register alpassvc_susp_type susp, snd_list_type snd_list)
} /* alpassvc_nn_fetch */
void alpassvc_ns_fetch(register alpassvc_susp_type susp, snd_list_type snd_list)
void alpassvc_ni_fetch(snd_susp_type a_susp, snd_list_type snd_list)
{
alpassvc_susp_type susp = (alpassvc_susp_type) a_susp;
int cnt = 0; /* how many samples computed */
sample_type delaysnd_x2_sample;
int togo;
int n;
sample_block_type out;
@@ -153,13 +167,24 @@ void alpassvc_ns_fetch(register alpassvc_susp_type susp, snd_list_type snd_list)
register long buflen_reg;
register sample_type * delayptr_reg;
register sample_type * endptr_reg;
register sample_type delaysnd_scale_reg = susp->delaysnd->scale;
register sample_block_values_type delaysnd_ptr_reg;
register double delaysnd_pHaSe_iNcR_rEg = susp->delaysnd_pHaSe_iNcR;
register double delaysnd_pHaSe_ReG;
register sample_type delaysnd_x1_sample_reg;
register sample_block_values_type input_ptr_reg;
falloc_sample_block(out, "alpassvc_ns_fetch");
falloc_sample_block(out, "alpassvc_ni_fetch");
out_ptr = out->samples;
snd_list->block = out;
/* make sure sounds are primed with first values */
if (!susp->started) {
susp->started = true;
susp_check_samples(delaysnd, delaysnd_ptr, delaysnd_cnt);
susp->delaysnd_x1_sample = (susp->delaysnd_cnt--, *(susp->delaysnd_ptr));
}
susp_check_samples(delaysnd, delaysnd_ptr, delaysnd_cnt);
delaysnd_x2_sample = *(susp->delaysnd_ptr);
while (cnt < max_sample_block_len) { /* outer loop */
/* first compute how many samples to generate in inner loop: */
/* don't overflow the output sample block: */
@@ -169,14 +194,11 @@ void alpassvc_ns_fetch(register alpassvc_susp_type susp, snd_list_type snd_list)
susp_check_term_samples(input, input_ptr, input_cnt);
togo = min(togo, susp->input_cnt);
/* don't run past the delaysnd input sample block: */
susp_check_samples(delaysnd, delaysnd_ptr, delaysnd_cnt);
togo = min(togo, susp->delaysnd_cnt);
/* don't run past terminate time */
if (susp->terminate_cnt != UNKNOWN &&
susp->terminate_cnt <= susp->susp.current + cnt + togo) {
togo = susp->terminate_cnt - (susp->susp.current + cnt);
if (togo < 0) togo = 0; /* avoids rounding errros */
if (togo == 0) break;
}
@@ -186,53 +208,64 @@ void alpassvc_ns_fetch(register alpassvc_susp_type susp, snd_list_type snd_list)
buflen_reg = susp->buflen;
delayptr_reg = susp->delayptr;
endptr_reg = susp->endptr;
delaysnd_ptr_reg = susp->delaysnd_ptr;
delaysnd_pHaSe_ReG = susp->delaysnd_pHaSe;
delaysnd_x1_sample_reg = susp->delaysnd_x1_sample;
input_ptr_reg = susp->input_ptr;
out_ptr_reg = out_ptr;
if (n) do { /* the inner sample computation loop */
register sample_type y, z, delaysamp;
register int delayi;
register sample_type *yptr;
register sample_type y, z, delaysamp;
register int delayi;
register sample_type *yptr;
if (delaysnd_pHaSe_ReG >= 1.0) {
delaysnd_x1_sample_reg = delaysnd_x2_sample;
/* pick up next sample as delaysnd_x2_sample: */
susp->delaysnd_ptr++;
susp_took(delaysnd_cnt, 1);
delaysnd_pHaSe_ReG -= 1.0;
susp_check_samples_break(delaysnd, delaysnd_ptr, delaysnd_cnt, delaysnd_x2_sample);
}
/* compute where to read y, we want y to be delay_snd samples
* after delay_ptr, where we write the new sample. First,
* conver from seconds to samples. Note: don't use actual sound_type
* names in comments! The translator isn't smart enough.
*/
delaysamp =
(delaysnd_x1_sample_reg * (1 - delaysnd_pHaSe_ReG) + delaysnd_x2_sample * delaysnd_pHaSe_ReG) * delay_scale_factor_reg;
delayi = (int) delaysamp; /* get integer part */
delaysamp = delaysamp - delayi; /* get phase */
yptr = delayptr_reg + buflen_reg - (delayi + 1);
if (yptr >= endptr_reg) yptr -= buflen_reg;
/* now get y, the out-put of the delay, using interpolation */
/* note that as phase increases, we use more of yptr[0] because
positive phase means longer buffer means read earlier sample */
y = (float) ((yptr[0] * delaysamp) + (yptr[1] * (1.0 - delaysamp)));
/* WARNING: no check to keep delaysamp in range, so
do this in LISP */
/* compute where to read y, we want y to be delay_snd samples
* after delay_ptr, where we write the new sample. First,
* conver from seconds to samples. Note: don't use actual sound_type
* names in comments! The translator isn't smart enough.
*/
delaysamp = (delaysnd_scale_reg * *delaysnd_ptr_reg++) * delay_scale_factor_reg;
delayi = (int) delaysamp; /* get integer part */
delaysamp = delaysamp - delayi; /* get phase */
yptr = delayptr_reg + buflen_reg - (delayi + 1);
if (yptr >= endptr_reg) yptr -= buflen_reg;
/* now get y, the out-put of the delay, using interpolation */
/* note that as phase increases, we use more of yptr[0] because
positive phase means longer buffer means read earlier sample */
y = (float) ((yptr[0] * delaysamp) + (yptr[1] * (1.0 - delaysamp)));
/* WARNING: no check to keep delaysamp in range, so do this in LISP */
*delayptr_reg++ = z = (sample_type) (feedback_reg * y + *input_ptr_reg++);
/* Time out to update the buffer:
* this is a tricky buffer: buffer[0] == buffer[bufflen]
* the logical length is bufflen, but the actual length
* is bufflen + 1 to allow for a repeated sample at the
* end. This allows for efficient interpolation.
*/
if (delayptr_reg > endptr_reg) {
delayptr_reg = susp->delaybuf;
*delayptr_reg++ = *endptr_reg;
}
*out_ptr_reg++ = (sample_type) (y - feedback_reg * z);;
*delayptr_reg++ = z = (sample_type) (feedback_reg * y + *input_ptr_reg++);
/* Time out to update the buffer:
* this is a tricky buffer: buffer[0] == buffer[bufflen]
* the logical length is bufflen, but the actual length
* is bufflen + 1 to allow for a repeated sample at the
* end. This allows for efficient interpolation.
*/
if (delayptr_reg > endptr_reg) {
delayptr_reg = susp->delaybuf;
*delayptr_reg++ = *endptr_reg;
}
*out_ptr_reg++ = (sample_type) (y - feedback_reg * z);
delaysnd_pHaSe_ReG += delaysnd_pHaSe_iNcR_rEg;
} while (--n); /* inner loop */
togo -= n;
susp->buflen = buflen_reg;
susp->delayptr = delayptr_reg;
/* using delaysnd_ptr_reg is a bad idea on RS/6000: */
susp->delaysnd_ptr += togo;
susp->delaysnd_pHaSe = delaysnd_pHaSe_ReG;
susp->delaysnd_x1_sample = delaysnd_x1_sample_reg;
/* using input_ptr_reg is a bad idea on RS/6000: */
susp->input_ptr += togo;
out_ptr += togo;
susp_took(input_cnt, togo);
susp_took(delaysnd_cnt, togo);
cnt += togo;
} /* outer loop */
@@ -243,14 +276,145 @@ void alpassvc_ns_fetch(register alpassvc_susp_type susp, snd_list_type snd_list)
snd_list->block_len = cnt;
susp->susp.current += cnt;
}
} /* alpassvc_ns_fetch */
} /* alpassvc_ni_fetch */
void alpassvc_toss_fetch(susp, snd_list)
register alpassvc_susp_type susp;
snd_list_type snd_list;
void alpassvc_nr_fetch(snd_susp_type a_susp, snd_list_type snd_list)
{
long final_count = susp->susp.toss_cnt;
alpassvc_susp_type susp = (alpassvc_susp_type) a_susp;
int cnt = 0; /* how many samples computed */
sample_type delaysnd_DeLtA;
sample_type delaysnd_val;
sample_type delaysnd_x2_sample;
int togo;
int n;
sample_block_type out;
register sample_block_values_type out_ptr;
register sample_block_values_type out_ptr_reg;
register float delay_scale_factor_reg;
register double feedback_reg;
register long buflen_reg;
register sample_type * delayptr_reg;
register sample_type * endptr_reg;
register sample_block_values_type input_ptr_reg;
falloc_sample_block(out, "alpassvc_nr_fetch");
out_ptr = out->samples;
snd_list->block = out;
/* make sure sounds are primed with first values */
if (!susp->started) {
susp->started = true;
susp->delaysnd_pHaSe = 1.0;
}
susp_check_samples(delaysnd, delaysnd_ptr, delaysnd_cnt);
delaysnd_x2_sample = *(susp->delaysnd_ptr);
while (cnt < max_sample_block_len) { /* outer loop */
/* first compute how many samples to generate in inner loop: */
/* don't overflow the output sample block: */
togo = max_sample_block_len - cnt;
/* don't run past the input input sample block: */
susp_check_term_samples(input, input_ptr, input_cnt);
togo = min(togo, susp->input_cnt);
/* grab next delaysnd_x2_sample when phase goes past 1.0; */
/* we use delaysnd_n (computed below) to avoid roundoff errors: */
if (susp->delaysnd_n <= 0) {
susp->delaysnd_x1_sample = delaysnd_x2_sample;
susp->delaysnd_ptr++;
susp_took(delaysnd_cnt, 1);
susp->delaysnd_pHaSe -= 1.0;
susp_check_samples(delaysnd, delaysnd_ptr, delaysnd_cnt);
delaysnd_x2_sample = *(susp->delaysnd_ptr);
/* delaysnd_n gets number of samples before phase exceeds 1.0: */
susp->delaysnd_n = (long) ((1.0 - susp->delaysnd_pHaSe) *
susp->output_per_delaysnd);
}
togo = min(togo, susp->delaysnd_n);
delaysnd_DeLtA = (sample_type) ((delaysnd_x2_sample - susp->delaysnd_x1_sample) * susp->delaysnd_pHaSe_iNcR);
delaysnd_val = (sample_type) (susp->delaysnd_x1_sample * (1.0 - susp->delaysnd_pHaSe) +
delaysnd_x2_sample * susp->delaysnd_pHaSe);
/* don't run past terminate time */
if (susp->terminate_cnt != UNKNOWN &&
susp->terminate_cnt <= susp->susp.current + cnt + togo) {
togo = susp->terminate_cnt - (susp->susp.current + cnt);
if (togo < 0) togo = 0; /* avoids rounding errros */
if (togo == 0) break;
}
n = togo;
delay_scale_factor_reg = susp->delay_scale_factor;
feedback_reg = susp->feedback;
buflen_reg = susp->buflen;
delayptr_reg = susp->delayptr;
endptr_reg = susp->endptr;
input_ptr_reg = susp->input_ptr;
out_ptr_reg = out_ptr;
if (n) do { /* the inner sample computation loop */
register sample_type y, z, delaysamp;
register int delayi;
register sample_type *yptr;
/* compute where to read y, we want y to be delay_snd samples
* after delay_ptr, where we write the new sample. First,
* conver from seconds to samples. Note: don't use actual sound_type
* names in comments! The translator isn't smart enough.
*/
delaysamp = delaysnd_val * delay_scale_factor_reg;
delayi = (int) delaysamp; /* get integer part */
delaysamp = delaysamp - delayi; /* get phase */
yptr = delayptr_reg + buflen_reg - (delayi + 1);
if (yptr >= endptr_reg) yptr -= buflen_reg;
/* now get y, the out-put of the delay, using interpolation */
/* note that as phase increases, we use more of yptr[0] because
positive phase means longer buffer means read earlier sample */
y = (float) ((yptr[0] * delaysamp) + (yptr[1] * (1.0 - delaysamp)));
/* WARNING: no check to keep delaysamp in range, so
do this in LISP */
*delayptr_reg++ = z = (sample_type) (feedback_reg * y + *input_ptr_reg++);
/* Time out to update the buffer:
* this is a tricky buffer: buffer[0] == buffer[bufflen]
* the logical length is bufflen, but the actual length
* is bufflen + 1 to allow for a repeated sample at the
* end. This allows for efficient interpolation.
*/
if (delayptr_reg > endptr_reg) {
delayptr_reg = susp->delaybuf;
*delayptr_reg++ = *endptr_reg;
}
*out_ptr_reg++ = (sample_type) (y - feedback_reg * z);
delaysnd_val += delaysnd_DeLtA;
} while (--n); /* inner loop */
susp->buflen = buflen_reg;
susp->delayptr = delayptr_reg;
/* using input_ptr_reg is a bad idea on RS/6000: */
susp->input_ptr += togo;
out_ptr += togo;
susp_took(input_cnt, togo);
susp->delaysnd_pHaSe += togo * susp->delaysnd_pHaSe_iNcR;
susp->delaysnd_n -= togo;
cnt += togo;
} /* outer loop */
/* test for termination */
if (togo == 0 && cnt == 0) {
snd_list_terminate(snd_list);
} else {
snd_list->block_len = cnt;
susp->susp.current += cnt;
}
} /* alpassvc_nr_fetch */
void alpassvc_toss_fetch(snd_susp_type a_susp, snd_list_type snd_list)
{
alpassvc_susp_type susp = (alpassvc_susp_type) a_susp;
time_type final_time = susp->susp.t0;
long n;
@@ -273,27 +437,31 @@ void alpassvc_toss_fetch(susp, snd_list)
susp->delaysnd_ptr += n;
susp_took(delaysnd_cnt, n);
susp->susp.fetch = susp->susp.keep_fetch;
(*(susp->susp.fetch))(susp, snd_list);
(*(susp->susp.fetch))(a_susp, snd_list);
}
void alpassvc_mark(alpassvc_susp_type susp)
void alpassvc_mark(snd_susp_type a_susp)
{
alpassvc_susp_type susp = (alpassvc_susp_type) a_susp;
sound_xlmark(susp->input);
sound_xlmark(susp->delaysnd);
}
void alpassvc_free(alpassvc_susp_type susp)
void alpassvc_free(snd_susp_type a_susp)
{
free(susp->delaybuf); sound_unref(susp->input);
alpassvc_susp_type susp = (alpassvc_susp_type) a_susp;
free(susp->delaybuf);
sound_unref(susp->input);
sound_unref(susp->delaysnd);
ffree_generic(susp, sizeof(alpassvc_susp_node), "alpassvc_free");
}
void alpassvc_print_tree(alpassvc_susp_type susp, int n)
void alpassvc_print_tree(snd_susp_type a_susp, int n)
{
alpassvc_susp_type susp = (alpassvc_susp_type) a_susp;
indent(n);
stdputstr("input:");
sound_print_tree_1(susp->input, n);
@@ -307,7 +475,7 @@ void alpassvc_print_tree(alpassvc_susp_type susp, int n)
sound_type snd_make_alpassvc(sound_type input, sound_type delaysnd, double feedback, double maxdelay)
{
register alpassvc_susp_type susp;
rate_type sr = max(input->sr, delaysnd->sr);
rate_type sr = input->sr;
time_type t0 = max(input->t0, delaysnd->t0);
int interp_desc = 0;
sample_type scale_factor = 1.0F;
@@ -327,12 +495,20 @@ sound_type snd_make_alpassvc(sound_type input, sound_type delaysnd, double feedb
susp->delayptr = susp->delaybuf;
susp->endptr = susp->delaybuf + susp->buflen;
/* make sure no sample rate is too high */
if (delaysnd->sr > sr) {
sound_unref(delaysnd);
snd_badsr();
}
/* select a susp fn based on sample rates */
interp_desc = (interp_desc << 2) + interp_style(input, sr);
interp_desc = (interp_desc << 2) + interp_style(delaysnd, sr);
switch (interp_desc) {
case INTERP_ns: /* handled below */
case INTERP_nn: susp->susp.fetch = alpassvc_nn_fetch; break;
case INTERP_ns: susp->susp.fetch = alpassvc_ns_fetch; break;
case INTERP_ni: susp->susp.fetch = alpassvc_ni_fetch; break;
case INTERP_nr: susp->susp.fetch = alpassvc_nr_fetch; break;
default: snd_badsr(); break;
}
@@ -345,8 +521,8 @@ sound_type snd_make_alpassvc(sound_type input, sound_type delaysnd, double feedb
/* how many samples to toss before t0: */
susp->susp.toss_cnt = (long) ((t0 - t0_min) * sr + 0.5);
if (susp->susp.toss_cnt > 0) {
susp->susp.keep_fetch = susp->susp.fetch;
susp->susp.fetch = alpassvc_toss_fetch;
susp->susp.keep_fetch = susp->susp.fetch;
susp->susp.fetch = alpassvc_toss_fetch;
}
/* initialize susp state */
@@ -357,11 +533,16 @@ sound_type snd_make_alpassvc(sound_type input, sound_type delaysnd, double feedb
susp->susp.print_tree = alpassvc_print_tree;
susp->susp.name = "alpassvc";
susp->susp.log_stop_cnt = UNKNOWN;
susp->started = false;
susp->susp.current = 0;
susp->input = input;
susp->input_cnt = 0;
susp->delaysnd = delaysnd;
susp->delaysnd_cnt = 0;
susp->delaysnd_pHaSe = 0.0;
susp->delaysnd_pHaSe_iNcR = delaysnd->sr / sr;
susp->delaysnd_n = 0;
susp->output_per_delaysnd = sr / delaysnd->sr;
return sound_create((snd_susp_type)susp, t0, sr, scale_factor);
}