/
synthesize.c
1617 lines (1355 loc) · 40.4 KB
/
synthesize.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
* Copyright (C) 2005 to 2014 by Jonathan Duddington
* email: jonsd@users.sourceforge.net
* Copyright (C) 2015-2017 Reece H. Dunn
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, see: <http://www.gnu.org/licenses/>.
*/
#include "config.h"
#include <ctype.h>
#include <errno.h>
#include <math.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <espeak-ng/espeak_ng.h>
#include <espeak-ng/speak_lib.h>
#include <espeak-ng/encoding.h>
#include "dictionary.h"
#include "intonation.h"
#include "mbrola.h"
#include "setlengths.h"
#include "synthdata.h"
#include "wavegen.h"
#include "phoneme.h"
#include "voice.h"
#include "synthesize.h"
#include "translate.h"
extern FILE *f_log;
static void SmoothSpect(void);
// list of phonemes in a clause
int n_phoneme_list = 0;
PHONEME_LIST phoneme_list[N_PHONEME_LIST+1];
SPEED_FACTORS speed;
static int last_pitch_cmd;
static int last_amp_cmd;
static frame_t *last_frame;
static int last_wcmdq;
static int pitch_length;
static int amp_length;
static int modn_flags;
static int fmt_amplitude = 0;
static int syllable_start;
static int syllable_end;
static int syllable_centre;
static voice_t *new_voice = NULL;
int n_soundicon_tab = N_SOUNDICON_SLOTS;
SOUND_ICON soundicon_tab[N_SOUNDICON_TAB];
#define RMS_GLOTTAL1 35 // vowel before glottal stop
#define RMS_START 28 // 28
#define VOWEL_FRONT_LENGTH 50
// a dummy phoneme_list entry which looks like a pause
static PHONEME_LIST next_pause;
const char *WordToString(unsigned int word)
{
// Convert a phoneme mnemonic word into a string
int ix;
static char buf[5];
for (ix = 0; ix < 4; ix++)
buf[ix] = word >> (ix*8);
buf[4] = 0;
return buf;
}
void SynthesizeInit()
{
last_pitch_cmd = 0;
last_amp_cmd = 0;
last_frame = NULL;
syllable_centre = -1;
// initialise next_pause, a dummy phoneme_list entry
next_pause.type = phPAUSE;
next_pause.newword = 0;
}
static void EndAmplitude(void)
{
if (amp_length > 0) {
if (wcmdq[last_amp_cmd][1] == 0)
wcmdq[last_amp_cmd][1] = amp_length;
amp_length = 0;
}
}
static void EndPitch(int voice_break)
{
// posssible end of pitch envelope, fill in the length
if ((pitch_length > 0) && (last_pitch_cmd >= 0)) {
if (wcmdq[last_pitch_cmd][1] == 0)
wcmdq[last_pitch_cmd][1] = pitch_length;
pitch_length = 0;
}
if (voice_break) {
last_wcmdq = -1;
last_frame = NULL;
syllable_end = wcmdq_tail;
SmoothSpect();
syllable_centre = -1;
memset(vowel_transition, 0, sizeof(vowel_transition));
}
}
static void DoAmplitude(int amp, unsigned char *amp_env)
{
intptr_t *q;
last_amp_cmd = wcmdq_tail;
amp_length = 0; // total length of vowel with this amplitude envelope
q = wcmdq[wcmdq_tail];
q[0] = WCMD_AMPLITUDE;
q[1] = 0; // fill in later from amp_length
q[2] = (intptr_t)amp_env;
q[3] = amp;
WcmdqInc();
}
static void DoPhonemeAlignment(char* pho, int type)
{
wcmdq[wcmdq_tail][0] = WCMD_PHONEME_ALIGNMENT;
wcmdq[wcmdq_tail][1] = pho;
wcmdq[wcmdq_tail][2] = type;
WcmdqInc();
}
static void DoPitch(unsigned char *env, int pitch1, int pitch2)
{
intptr_t *q;
EndPitch(0);
if (pitch1 == 255) {
// pitch was not set
pitch1 = 55;
pitch2 = 76;
env = envelope_data[PITCHfall];
}
last_pitch_cmd = wcmdq_tail;
pitch_length = 0; // total length of spect with this pitch envelope
if (pitch2 < 0)
pitch2 = 0;
q = wcmdq[wcmdq_tail];
q[0] = WCMD_PITCH;
q[1] = 0; // length, fill in later from pitch_length
q[2] = (intptr_t)env;
q[3] = (pitch1 << 16) + pitch2;
WcmdqInc();
}
int PauseLength(int pause, int control)
{
unsigned int len;
if (control == 0) {
if (pause >= 200)
len = (pause * speed.clause_pause_factor)/256;
else
len = (pause * speed.pause_factor)/256;
} else
len = (pause * speed.wav_factor)/256;
if (len < speed.min_pause)
len = speed.min_pause; // mS, limit the amount to which pauses can be shortened
return len;
}
static void DoPause(int length, int control)
{
// length in nominal mS
// control = 1, less shortening at fast speeds
unsigned int len;
int srate2;
if (length == 0)
len = 0;
else {
len = PauseLength(length, control);
if (len < 90000)
len = (len * samplerate) / 1000; // convert from mS to number of samples
else {
srate2 = samplerate / 25; // avoid overflow
len = (len * srate2) / 40;
}
}
EndPitch(1);
wcmdq[wcmdq_tail][0] = WCMD_PAUSE;
wcmdq[wcmdq_tail][1] = len;
WcmdqInc();
last_frame = NULL;
if (fmt_amplitude != 0) {
wcmdq[wcmdq_tail][0] = WCMD_FMT_AMPLITUDE;
wcmdq[wcmdq_tail][1] = fmt_amplitude = 0;
WcmdqInc();
}
}
extern int seq_len_adjust; // temporary fix to advance the start point for playing the wav sample
static int DoSample2(int index, int which, int std_length, int control, int length_mod, int amp)
{
int length;
int wav_length;
int wav_scale;
int min_length;
int x;
int len4;
intptr_t *q;
unsigned char *p;
index = index & 0x7fffff;
p = &wavefile_data[index];
wav_scale = p[2];
wav_length = (p[1] * 256);
wav_length += p[0]; // length in bytes
if (wav_length == 0)
return 0;
min_length = speed.min_sample_len;
if (wav_scale == 0)
min_length *= 2; // 16 bit samples
if (std_length > 0) {
std_length = (std_length * samplerate)/1000;
if (wav_scale == 0)
std_length *= 2;
x = (min_length * std_length)/wav_length;
if (x > min_length)
min_length = x;
} else {
// no length specified, use the length of the stored sound
std_length = wav_length;
}
if (length_mod > 0)
std_length = (std_length * length_mod)/256;
length = (std_length * speed.wav_factor)/256;
if (control & pd_DONTLENGTHEN) {
// this option is used for Stops, with short noise bursts.
// Don't change their length much.
if (length > std_length) {
// don't let length exceed std_length
length = std_length;
}
}
if (length < min_length)
length = min_length;
if (wav_scale == 0) {
// 16 bit samples
length /= 2;
wav_length /= 2;
}
if (amp < 0)
return length;
len4 = wav_length / 4;
index += 4;
if (which & 0x100) {
// mix this with synthesised wave
last_wcmdq = wcmdq_tail;
q = wcmdq[wcmdq_tail];
q[0] = WCMD_WAVE2;
q[1] = length | (wav_length << 16); // length in samples
q[2] = (intptr_t)(&wavefile_data[index]);
q[3] = wav_scale + (amp << 8);
WcmdqInc();
return length;
}
if (length > wav_length) {
x = len4*3;
length -= x;
} else {
x = length;
length = 0;
}
last_wcmdq = wcmdq_tail;
q = wcmdq[wcmdq_tail];
q[0] = WCMD_WAVE;
q[1] = x; // length in samples
q[2] = (intptr_t)(&wavefile_data[index]);
q[3] = wav_scale + (amp << 8);
WcmdqInc();
while (length > len4*3) {
x = len4;
if (wav_scale == 0)
x *= 2;
last_wcmdq = wcmdq_tail;
q = wcmdq[wcmdq_tail];
q[0] = WCMD_WAVE;
q[1] = len4*2; // length in samples
q[2] = (intptr_t)(&wavefile_data[index+x]);
q[3] = wav_scale + (amp << 8);
WcmdqInc();
length -= len4*2;
}
if (length > 0) {
x = wav_length - length;
if (wav_scale == 0)
x *= 2;
last_wcmdq = wcmdq_tail;
q = wcmdq[wcmdq_tail];
q[0] = WCMD_WAVE;
q[1] = length; // length in samples
q[2] = (intptr_t)(&wavefile_data[index+x]);
q[3] = wav_scale + (amp << 8);
WcmdqInc();
}
return length;
}
int DoSample3(PHONEME_DATA *phdata, int length_mod, int amp)
{
int amp2;
int len;
EndPitch(1);
if (amp == -1) {
// just get the length, don't produce sound
amp2 = amp;
} else {
amp2 = phdata->sound_param[pd_WAV];
if (amp2 == 0)
amp2 = 100;
amp2 = (amp2 * 32)/100;
}
seq_len_adjust = 0;
if (phdata->sound_addr[pd_WAV] == 0)
len = 0;
else
len = DoSample2(phdata->sound_addr[pd_WAV], 2, phdata->pd_param[pd_LENGTHMOD]*2, phdata->pd_control, length_mod, amp2);
last_frame = NULL;
return len;
}
static frame_t *AllocFrame()
{
// Allocate a temporary spectrum frame for the wavegen queue. Use a pool which is big
// enough to use a round-robin without checks.
// Only needed for modifying spectra for blending to consonants
#define N_FRAME_POOL N_WCMDQ
static int ix = 0;
static frame_t frame_pool[N_FRAME_POOL];
ix++;
if (ix >= N_FRAME_POOL)
ix = 0;
return &frame_pool[ix];
}
static void set_frame_rms(frame_t *fr, int new_rms)
{
// Each frame includes its RMS amplitude value, so to set a new
// RMS just adjust the formant amplitudes by the appropriate ratio
int x;
int h;
int ix;
static const short sqrt_tab[200] = {
0, 64, 90, 110, 128, 143, 156, 169, 181, 192, 202, 212, 221, 230, 239, 247,
256, 263, 271, 278, 286, 293, 300, 306, 313, 320, 326, 332, 338, 344, 350, 356,
362, 367, 373, 378, 384, 389, 394, 399, 404, 409, 414, 419, 424, 429, 434, 438,
443, 448, 452, 457, 461, 465, 470, 474, 478, 483, 487, 491, 495, 499, 503, 507,
512, 515, 519, 523, 527, 531, 535, 539, 543, 546, 550, 554, 557, 561, 565, 568,
572, 576, 579, 583, 586, 590, 593, 596, 600, 603, 607, 610, 613, 617, 620, 623,
627, 630, 633, 636, 640, 643, 646, 649, 652, 655, 658, 662, 665, 668, 671, 674,
677, 680, 683, 686, 689, 692, 695, 698, 701, 704, 706, 709, 712, 715, 718, 721,
724, 726, 729, 732, 735, 738, 740, 743, 746, 749, 751, 754, 757, 759, 762, 765,
768, 770, 773, 775, 778, 781, 783, 786, 789, 791, 794, 796, 799, 801, 804, 807,
809, 812, 814, 817, 819, 822, 824, 827, 829, 832, 834, 836, 839, 841, 844, 846,
849, 851, 853, 856, 858, 861, 863, 865, 868, 870, 872, 875, 877, 879, 882, 884,
886, 889, 891, 893, 896, 898, 900, 902
};
if (voice->klattv[0]) {
if (new_rms == -1)
fr->klattp[KLATT_AV] = 50;
return;
}
if (fr->rms == 0) return; // check for divide by zero
x = (new_rms * 64)/fr->rms;
if (x >= 200) x = 199;
x = sqrt_tab[x]; // sqrt(new_rms/fr->rms)*0x200;
for (ix = 0; ix < 8; ix++) {
h = fr->fheight[ix] * x;
fr->fheight[ix] = h/0x200;
}
}
static void formants_reduce_hf(frame_t *fr, int level)
{
// change height of peaks 2 to 8, percentage
int ix;
int x;
if (voice->klattv[0])
return;
for (ix = 2; ix < 8; ix++) {
x = fr->fheight[ix] * level;
fr->fheight[ix] = x/100;
}
}
static frame_t *CopyFrame(frame_t *frame1, int copy)
{
// create a copy of the specified frame in temporary buffer
frame_t *frame2;
if ((copy == 0) && (frame1->frflags & FRFLAG_COPIED)) {
// this frame has already been copied in temporary rw memory
return frame1;
}
frame2 = AllocFrame();
if (frame2 != NULL) {
memcpy(frame2, frame1, sizeof(frame_t));
frame2->length = 0;
frame2->frflags |= FRFLAG_COPIED;
}
return frame2;
}
static frame_t *DuplicateLastFrame(frameref_t *seq, int n_frames, int length)
{
frame_t *fr;
seq[n_frames-1].length = length;
fr = CopyFrame(seq[n_frames-1].frame, 1);
seq[n_frames].frame = fr;
seq[n_frames].length = 0;
return fr;
}
static void AdjustFormants(frame_t *fr, int target, int min, int max, int f1_adj, int f3_adj, int hf_reduce, int flags)
{
int x;
target = (target * voice->formant_factor)/256;
x = (target - fr->ffreq[2]) / 2;
if (x > max) x = max;
if (x < min) x = min;
fr->ffreq[2] += x;
fr->ffreq[3] += f3_adj;
if (flags & 0x20)
f3_adj = -f3_adj; // reverse direction for f4,f5 change
fr->ffreq[4] += f3_adj;
fr->ffreq[5] += f3_adj;
if (f1_adj == 1) {
x = (235 - fr->ffreq[1]);
if (x < -100) x = -100;
if (x > -60) x = -60;
fr->ffreq[1] += x;
}
if (f1_adj == 2) {
x = (235 - fr->ffreq[1]);
if (x < -300) x = -300;
if (x > -150) x = -150;
fr->ffreq[1] += x;
fr->ffreq[0] += x;
}
if (f1_adj == 3) {
x = (100 - fr->ffreq[1]);
if (x < -400) x = -400;
if (x > -300) x = -400;
fr->ffreq[1] += x;
fr->ffreq[0] += x;
}
formants_reduce_hf(fr, hf_reduce);
}
static int VowelCloseness(frame_t *fr)
{
// return a value 0-3 depending on the vowel's f1
int f1;
if ((f1 = fr->ffreq[1]) < 300)
return 3;
if (f1 < 400)
return 2;
if (f1 < 500)
return 1;
return 0;
}
int FormantTransition2(frameref_t *seq, int *n_frames, unsigned int data1, unsigned int data2, PHONEME_TAB *other_ph, int which)
{
int ix;
int formant;
int next_rms;
int len;
int rms;
int f1;
int f2;
int f2_min;
int f2_max;
int f3_adj;
int f3_amp;
int flags;
int vcolour;
#define N_VCOLOUR 2
// percentage change for each formant in 256ths
static short vcolouring[N_VCOLOUR][5] = {
{ 243, 272, 256, 256, 256 }, // palatal consonant follows
{ 256, 256, 240, 240, 240 }, // retroflex
};
frame_t *fr = NULL;
if (*n_frames < 2)
return 0;
len = (data1 & 0x3f) * 2;
rms = (data1 >> 6) & 0x3f;
flags = (data1 >> 12);
f2 = (data2 & 0x3f) * 50;
f2_min = (((data2 >> 6) & 0x1f) - 15) * 50;
f2_max = (((data2 >> 11) & 0x1f) - 15) * 50;
f3_adj = (((data2 >> 16) & 0x1f) - 15) * 50;
f3_amp = ((data2 >> 21) & 0x1f) * 8;
f1 = ((data2 >> 26) & 0x7);
vcolour = (data2 >> 29);
if ((other_ph != NULL) && (other_ph->mnemonic == '?'))
flags |= 8;
if (which == 1) {
// entry to vowel
fr = CopyFrame(seq[0].frame, 0);
seq[0].frame = fr;
seq[0].length = VOWEL_FRONT_LENGTH;
if (len > 0)
seq[0].length = len;
seq[0].frflags |= FRFLAG_LEN_MOD2; // reduce length modification
fr->frflags |= FRFLAG_LEN_MOD2;
next_rms = seq[1].frame->rms;
if (voice->klattv[0])
fr->klattp[KLATT_AV] = seq[1].frame->klattp[KLATT_AV] - 4;
if (f2 != 0) {
if (rms & 0x20)
set_frame_rms(fr, (next_rms * (rms & 0x1f))/30);
AdjustFormants(fr, f2, f2_min, f2_max, f1, f3_adj, f3_amp, flags);
if ((rms & 0x20) == 0)
set_frame_rms(fr, rms*2);
} else {
if (flags & 8)
set_frame_rms(fr, (next_rms*24)/32);
else
set_frame_rms(fr, RMS_START);
}
if (flags & 8)
modn_flags = 0x800 + (VowelCloseness(fr) << 8);
} else {
// exit from vowel
rms = rms*2;
if ((f2 != 0) || (flags != 0)) {
if (flags & 8) {
fr = CopyFrame(seq[*n_frames-1].frame, 0);
seq[*n_frames-1].frame = fr;
rms = RMS_GLOTTAL1;
// degree of glottal-stop effect depends on closeness of vowel (indicated by f1 freq)
modn_flags = 0x400 + (VowelCloseness(fr) << 8);
} else {
fr = DuplicateLastFrame(seq, (*n_frames)++, len);
if (len > 36)
seq_len_adjust += (len - 36);
if (f2 != 0)
AdjustFormants(fr, f2, f2_min, f2_max, f1, f3_adj, f3_amp, flags);
}
set_frame_rms(fr, rms);
if ((vcolour > 0) && (vcolour <= N_VCOLOUR)) {
for (ix = 0; ix < *n_frames; ix++) {
fr = CopyFrame(seq[ix].frame, 0);
seq[ix].frame = fr;
for (formant = 1; formant <= 5; formant++) {
int x;
x = fr->ffreq[formant] * vcolouring[vcolour-1][formant-1];
fr->ffreq[formant] = x / 256;
}
}
}
}
}
if (fr != NULL) {
if (flags & 4)
fr->frflags |= FRFLAG_FORMANT_RATE;
if (flags & 2)
fr->frflags |= FRFLAG_BREAK; // don't merge with next frame
}
if (flags & 0x40)
DoPause(20, 0); // add a short pause after the consonant
if (flags & 16)
return len;
return 0;
}
static void SmoothSpect(void)
{
// Limit the rate of frequence change of formants, to reduce chirping
intptr_t *q;
frame_t *frame;
frame_t *frame2;
frame_t *frame1;
frame_t *frame_centre;
int ix;
int len;
int pk;
bool modified;
int allowed;
int diff;
if (syllable_start == syllable_end)
return;
if ((syllable_centre < 0) || (syllable_centre == syllable_start)) {
syllable_start = syllable_end;
return;
}
q = wcmdq[syllable_centre];
frame_centre = (frame_t *)q[2];
// backwards
ix = syllable_centre -1;
frame = frame2 = frame_centre;
for (;;) {
if (ix < 0) ix = N_WCMDQ-1;
q = wcmdq[ix];
if (q[0] == WCMD_PAUSE || q[0] == WCMD_WAVE)
break;
if (q[0] <= WCMD_SPECT2) {
len = q[1] & 0xffff;
frame1 = (frame_t *)q[3];
if (frame1 == frame) {
q[3] = (intptr_t)frame2;
frame1 = frame2;
} else
break; // doesn't follow on from previous frame
frame = frame2 = (frame_t *)q[2];
modified = false;
if (frame->frflags & FRFLAG_BREAK)
break;
if (frame->frflags & FRFLAG_FORMANT_RATE)
len = (len * 12)/10; // allow slightly greater rate of change for this frame (was 12/10)
for (pk = 0; pk < 6; pk++) {
int f1, f2;
if ((frame->frflags & FRFLAG_BREAK_LF) && (pk < 3))
continue;
f1 = frame1->ffreq[pk];
f2 = frame->ffreq[pk];
// backwards
if ((diff = f2 - f1) > 0)
allowed = f1*2 + f2;
else
allowed = f1 + f2*2;
// the allowed change is specified as percentage (%*10) of the frequency
// take "frequency" as 1/3 from the lower freq
allowed = (allowed * formant_rate[pk])/3000;
allowed = (allowed * len)/256;
if (diff > allowed) {
if (modified == false) {
frame2 = CopyFrame(frame, 0);
modified = true;
}
frame2->ffreq[pk] = frame1->ffreq[pk] + allowed;
q[2] = (intptr_t)frame2;
} else if (diff < -allowed) {
if (modified == false) {
frame2 = CopyFrame(frame, 0);
modified = true;
}
frame2->ffreq[pk] = frame1->ffreq[pk] - allowed;
q[2] = (intptr_t)frame2;
}
}
}
if (ix == syllable_start)
break;
ix--;
}
// forwards
ix = syllable_centre;
frame = NULL;
for (;;) {
q = wcmdq[ix];
if (q[0] == WCMD_PAUSE || q[0] == WCMD_WAVE)
break;
if (q[0] <= WCMD_SPECT2) {
len = q[1] & 0xffff;
frame1 = (frame_t *)q[2];
if (frame != NULL) {
if (frame1 == frame) {
q[2] = (intptr_t)frame2;
frame1 = frame2;
} else
break; // doesn't follow on from previous frame
}
frame = frame2 = (frame_t *)q[3];
modified = false;
if (frame1->frflags & FRFLAG_BREAK)
break;
if (frame1->frflags & FRFLAG_FORMANT_RATE)
len = (len *6)/5; // allow slightly greater rate of change for this frame
for (pk = 0; pk < 6; pk++) {
int f1, f2;
f1 = frame1->ffreq[pk];
f2 = frame->ffreq[pk];
// forwards
if ((diff = f2 - f1) > 0)
allowed = f1*2 + f2;
else
allowed = f1 + f2*2;
allowed = (allowed * formant_rate[pk])/3000;
allowed = (allowed * len)/256;
if (diff > allowed) {
if (modified == false) {
frame2 = CopyFrame(frame, 0);
modified = true;
}
frame2->ffreq[pk] = frame1->ffreq[pk] + allowed;
q[3] = (intptr_t)frame2;
} else if (diff < -allowed) {
if (modified == false) {
frame2 = CopyFrame(frame, 0);
modified = true;
}
frame2->ffreq[pk] = frame1->ffreq[pk] - allowed;
q[3] = (intptr_t)frame2;
}
}
}
ix++;
if (ix >= N_WCMDQ) ix = 0;
if (ix == syllable_end)
break;
}
syllable_start = syllable_end;
}
static void StartSyllable(void)
{
// start of syllable, if not already started
if (syllable_end == syllable_start)
syllable_end = wcmdq_tail;
}
int DoSpect2(PHONEME_TAB *this_ph, int which, FMT_PARAMS *fmt_params, PHONEME_LIST *plist, int modulation)
{
// which: 0 not a vowel, 1 start of vowel, 2 body and end of vowel
// length_mod: 256 = 100%
// modulation: -1 = don't write to wcmdq
int n_frames;
frameref_t *frames;
int frameix;
frame_t *frame1;
frame_t *frame2;
frame_t *fr;
int ix;
intptr_t *q;
int len;
int frame_length;
int length_factor;
int length_mod;
int length_sum;
int length_min;
int total_len = 0;
static int wave_flag = 0;
int wcmd_spect = WCMD_SPECT;
int frame_lengths[N_SEQ_FRAMES];
if (fmt_params->fmt_addr == 0)
return 0;
length_mod = plist->length;
if (length_mod == 0) length_mod = 256;
length_min = (samplerate/70); // greater than one cycle at low pitch (Hz)
if (which == 2) {
if ((translator->langopts.param[LOPT_LONG_VOWEL_THRESHOLD] > 0) && ((this_ph->std_length >= translator->langopts.param[LOPT_LONG_VOWEL_THRESHOLD]) || (plist->synthflags & SFLAG_LENGTHEN) || (this_ph->phflags & phLONG)))
length_min *= 2; // ensure long vowels are longer
}
if (which == 1) {
// limit the shortening of sonorants before shortened (eg. unstressed vowels)
if ((this_ph->type == phLIQUID) || (plist[-1].type == phLIQUID) || (plist[-1].type == phNASAL)) {
if (length_mod < (len = translator->langopts.param[LOPT_SONORANT_MIN]))
length_mod = len;
}
}
modn_flags = 0;
frames = LookupSpect(this_ph, which, fmt_params, &n_frames, plist);
if (frames == NULL)
return 0; // not found
if (fmt_params->fmt_amp != fmt_amplitude) {
// an amplitude adjustment is specified for this sequence
q = wcmdq[wcmdq_tail];
q[0] = WCMD_FMT_AMPLITUDE;
q[1] = fmt_amplitude = fmt_params->fmt_amp;
WcmdqInc();
}
frame1 = frames[0].frame;
if (voice->klattv[0])
wcmd_spect = WCMD_KLATT;
wavefile_ix = fmt_params->wav_addr;
if (fmt_params->wav_amp == 0)
wavefile_amp = 32;
else
wavefile_amp = (fmt_params->wav_amp * 32)/100;
if (wavefile_ix == 0) {
if (wave_flag) {
// cancel any wavefile that was playing previously
wcmd_spect = WCMD_SPECT2;
if (voice->klattv[0])
wcmd_spect = WCMD_KLATT2;
wave_flag = 0;
} else {
wcmd_spect = WCMD_SPECT;
if (voice->klattv[0])
wcmd_spect = WCMD_KLATT;
}
}
if (last_frame != NULL) {
if (((last_frame->length < 2) || (last_frame->frflags & FRFLAG_VOWEL_CENTRE))
&& !(last_frame->frflags & FRFLAG_BREAK)) {
// last frame of previous sequence was zero-length, replace with first of this sequence
wcmdq[last_wcmdq][3] = (intptr_t)frame1;
if (last_frame->frflags & FRFLAG_BREAK_LF) {
// but flag indicates keep HF peaks in last segment
fr = CopyFrame(frame1, 1);
for (ix = 3; ix < 8; ix++) {
if (ix < 7)
fr->ffreq[ix] = last_frame->ffreq[ix];
fr->fheight[ix] = last_frame->fheight[ix];
}
wcmdq[last_wcmdq][3] = (intptr_t)fr;
}
}
}
if ((this_ph->type == phVOWEL) && (which == 2)) {
SmoothSpect(); // process previous syllable
// remember the point in the output queue of the centre of the vowel
syllable_centre = wcmdq_tail;
}
length_sum = 0;
for (frameix = 1; frameix < n_frames; frameix++) {
length_factor = length_mod;
if (frames[frameix-1].frflags & FRFLAG_LEN_MOD) // reduce effect of length mod
length_factor = (length_mod*(256-speed.lenmod_factor) + 256*speed.lenmod_factor)/256;
else if (frames[frameix-1].frflags & FRFLAG_LEN_MOD2) // reduce effect of length mod, used for the start of a vowel
length_factor = (length_mod*(256-speed.lenmod2_factor) + 256*speed.lenmod2_factor)/256;
frame_length = frames[frameix-1].length;
len = (frame_length * samplerate)/1000;
len = (len * length_factor)/256;
length_sum += len;
frame_lengths[frameix] = len;
}
if ((length_sum > 0) && (length_sum < length_min)) {
// lengthen, so that the sequence is greater than one cycle at low pitch
for (frameix = 1; frameix < n_frames; frameix++)
frame_lengths[frameix] = (frame_lengths[frameix] * length_min) / length_sum;
}
for (frameix = 1; frameix < n_frames; frameix++) {
frame2 = frames[frameix].frame;
if ((fmt_params->wav_addr != 0) && ((frame1->frflags & FRFLAG_DEFER_WAV) == 0)) {
// there is a wave file to play along with this synthesis
seq_len_adjust = 0;
DoSample2(fmt_params->wav_addr, which+0x100, 0, fmt_params->fmt_control, 0, wavefile_amp);
wave_flag = 1;
wavefile_ix = 0;
fmt_params->wav_addr = 0;
}
if (modulation >= 0) {
if (frame1->frflags & FRFLAG_MODULATE)
modulation = 6;
if ((frameix == n_frames-1) && (modn_flags & 0xf00))
modulation |= modn_flags; // before or after a glottal stop