1 /*************************************************************************/
2 /* */
3 /* Language Technologies Institute */
4 /* Carnegie Mellon University */
5 /* Copyright (c) 2001 */
6 /* All Rights Reserved. */
7 /* */
8 /* Permission is hereby granted, free of charge, to use and distribute */
9 /* this software and its documentation without restriction, including */
10 /* without limitation the rights to use, copy, modify, merge, publish, */
11 /* distribute, sublicense, and/or sell copies of this work, and to */
12 /* permit persons to whom this work is furnished to do so, subject to */
13 /* the following conditions: */
14 /* 1. The code must retain the above copyright notice, this list of */
15 /* conditions and the following disclaimer. */
16 /* 2. Any modifications must be clearly marked as such. */
17 /* 3. Original authors' names are not deleted. */
18 /* 4. The authors' names are not used to endorse or promote products */
19 /* derived from this software without specific prior written */
20 /* permission. */
21 /* */
22 /* CARNEGIE MELLON UNIVERSITY AND THE CONTRIBUTORS TO THIS WORK */
23 /* DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING */
24 /* ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT */
25 /* SHALL CARNEGIE MELLON UNIVERSITY NOR THE CONTRIBUTORS BE LIABLE */
26 /* FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES */
27 /* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN */
28 /* AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, */
29 /* ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF */
30 /* THIS SOFTWARE. */
31 /* */
32 /*************************************************************************/
33 /* Author: Alan W Black (awb@cs.cmu.edu) */
34 /* Date: January 2001 */
35 /*************************************************************************/
36 /* */
37 /* Diphone specific functions */
38 /* */
39 /*************************************************************************/
40
41 #include "cst_math.h"
42 #include "cst_hrg.h"
43 #include "cst_utt_utils.h"
44 #include "cst_wave.h"
45 #include "cst_track.h"
46 #include "cst_diphone.h"
47 #include "cst_units.h"
48 #include "cst_sigpr.h"
49
50 CST_VAL_REGISTER_TYPE_NODEL(diphone_db,cst_diphone_db)
51
52 static int get_diphone_entry(cst_diphone_db *udb,const char *name);
53 static int get_diphone_entry_bsearch(const cst_diphone_entry *entries,
54 int start, int end,
55 const char *key);
56
diphone_synth(cst_utterance * utt)57 cst_utterance* diphone_synth(cst_utterance *utt)
58 {
59 get_diphone_units(utt);
60 join_units(utt);
61 return utt;
62 }
63
get_diphone_units(cst_utterance * utt)64 cst_utterance *get_diphone_units(cst_utterance *utt)
65 {
66 /* Select units from db with times etc */
67 cst_relation *units;
68 cst_item *s0, *s1, *u;
69 float end0,end1;
70 char diphone_name[22];
71 cst_diphone_db *udb;
72 int unit_entry = -1;
73
74 udb = val_diphone_db(utt_feat_val(utt,"diphone_db"));
75 utt_set_feat(utt,"sts_list",sts_list_val(udb->sts));
76
77 units = utt_relation_create(utt,"Unit");
78
79 for (s0=relation_head(utt_relation(utt,"Segment"));
80 s0 && item_next(s0); s0=s1)
81 {
82 s1 = item_next(s0);
83 unit_entry = -1;
84 if (cst_streq("-",ffeature_string(s0,"ph_vc")) &&
85 cst_streq("-",ffeature_string(s0,"R:SylStructure.n.ph_vc")))
86 { /* Might have consonant cluster diphones */
87 cst_sprintf(diphone_name,
88 "%.10s_-_%.10s",
89 item_name(s0),
90 item_name(s1));
91 unit_entry = get_diphone_entry(udb,diphone_name);
92 }
93
94 if (unit_entry == -1) /* no consonant cluster diphone */
95 {
96 cst_sprintf(diphone_name,
97 "%.10s-%.10s",
98 item_name(s0),
99 item_name(s1));
100
101 unit_entry = get_diphone_entry(udb,diphone_name);
102 }
103
104 if (unit_entry == -1)
105 {
106 cst_errmsg("flite: udb failed to find entry for: %s\n",
107 diphone_name);
108 unit_entry = 0;
109 }
110
111 /* first half of diphone */
112 u = relation_append(units,NULL);
113 item_add_daughter(s0,u);
114 item_set_string(u,"name",diphone_name);
115 end0 = item_feat_float(s0,"end");
116 item_set_int(u,"target_end", (int)(end0*udb->sts->sample_rate));
117 item_set_int(u,"unit_entry",unit_entry);
118 item_set_int(u,"unit_start",udb->diphones[unit_entry].start_pm);
119 item_set_int(u,"unit_end",
120 udb->diphones[unit_entry].start_pm +
121 udb->diphones[unit_entry].pb_pm);
122 /* second half of diphone */
123 u = relation_append(units,NULL);
124 item_add_daughter(s1,u);
125 item_set_string(u,"name",diphone_name);
126 end1 = item_feat_float(s1,"end");
127 item_set_int(u,"target_end",(int)(((end0+end1)/2.0)*udb->sts->sample_rate));
128 item_set_int(u,"unit_entry",unit_entry);
129 item_set_int(u,"unit_start",
130 udb->diphones[unit_entry].start_pm +
131 udb->diphones[unit_entry].pb_pm);
132 item_set_int(u,"unit_end",
133 udb->diphones[unit_entry].start_pm +
134 udb->diphones[unit_entry].pb_pm+
135 udb->diphones[unit_entry].end_pm);
136 }
137
138 return utt;
139 }
140
get_diphone_entry(cst_diphone_db * udb,const char * name)141 static int get_diphone_entry(cst_diphone_db *udb, const char *name)
142 {
143 return get_diphone_entry_bsearch(udb->diphones,0,
144 udb->num_entries,
145 name);
146 }
147
get_diphone_entry_bsearch(const cst_diphone_entry * entries,int start,int end,const char * key)148 static int get_diphone_entry_bsearch(const cst_diphone_entry *entries,
149 int start, int end,
150 const char *key)
151 {
152 int mid,c;
153
154 while (start < end)
155 {
156 mid = (start+end)/2;
157
158 c = strcmp(entries[mid].name,key);
159
160 if (c == 0)
161 return mid;
162 else if (c > 0)
163 end = mid;
164 else
165 start = mid + 1;
166 }
167
168 return -1; /* can't find it */
169 }
170
171