Lang=hr, transliterate Cyrillic characters to Latin using TranslateChar().

TranslateChar() can now substitute a character by two replacement characters.
Lang=cs, more difference between long and short vowels, less length differnence beteeen stressed and unstressed syllables.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@7 d46cf337-b52f-0410-862d-fd96e6ae7743
This commit is contained in:
jonsd
2007-05-10 15:11:22 +00:00
parent 4a11084bd5
commit e24c5b339d
52 changed files with 448 additions and 198 deletions
+47 -22
View File
@@ -876,6 +876,7 @@ void Translator::SetWordStress(char *output, unsigned int dictionary_flags, int
int final_ph;
int mnem;
int post_tonic;
int opt_length;
unsigned char vowel_stress[N_WORD_PHONEMES/2];
char syllable_type[N_WORD_PHONEMES/2];
@@ -1143,19 +1144,20 @@ void Translator::SetWordStress(char *output, unsigned int dictionary_flags, int
tonic = langopts.unstressed_wd2; /* more than one syllable, used secondary stress as the main stress */
}
max_stress = 0;
max_stress_posn = 0;
for(v=1; v<vowel_count; v++)
{
if(vowel_stress[v] >= max_stress)
{
max_stress = vowel_stress[v];
max_stress_posn = v;
}
}
if(tonic >= 0)
{
/* find position of highest stress, and replace it by 'tonic' */
max_stress = 0;
max_stress_posn = 0;
for(v=1; v<vowel_count; v++)
{
if(vowel_stress[v] >= max_stress)
{
max_stress = vowel_stress[v];
max_stress_posn = v;
}
}
/* don't disturb an explicitly set stress by 'unstress-at-end' flag */
if((tonic > max_stress) || (max_stress <= 4))
@@ -1303,20 +1305,36 @@ void Translator::SetWordStress(char *output, unsigned int dictionary_flags, int
break;
}
if(langopts.param[LOPT_IT_LENGTHEN] && (*p == phonLENGTHEN))
{
// remove lengthen indicator from non-stressed syllables
if(v_stress < 4)
p++;
else
if((langopts.param[LOPT_IT_LENGTHEN]==2) && (v != (vowel_count - 2)))
p++; // LANG=Italian, remove lengthen indicator from non-penultimate syllables
}
if(vowel_stress[v] > max_stress)
{
max_stress = vowel_stress[v];
}
if((*p == phonLENGTHEN) && ((opt_length = langopts.param[LOPT_IT_LENGTHEN]) != 0))
{
// remove lengthen indicator from non-stressed syllables
int shorten=0;
if(opt_length & 0x10)
{
// only allow lengthen indicator on the highest stress syllable in the word
if(v != max_stress_posn)
shorten = 1;
}
else
if(v_stress < 4)
{
// only allow lengthen indicator if stress >= 4.
shorten = 1;
}
if(((opt_length & 0xf)==2) && (v != (vowel_count - 2)))
shorten = 1; // LANG=Italian, remove lengthen indicator from non-penultimate syllables
if(shorten)
p++;
}
v++;
}
@@ -1553,6 +1571,7 @@ void Translator::MatchRule(char *word[], const char *group, char *rule, MatchRec
int letter_group;
int distance_right;
int distance_left;
int lg_pts;
MatchRecord match;
static MatchRecord best;
@@ -1682,7 +1701,10 @@ void Translator::MatchRule(char *word[], const char *group, char *rule, MatchRec
letter_group = *rule++ - 'A';
if(IsLetter(letter_w,letter_group))
{
match.points += (20-distance_right);
lg_pts = 20;
if(letter_group==2)
lg_pts = 19; // fewer points for C, general consonant
match.points += (lg_pts-distance_right);
post_ptr += letter_xbytes;
}
else
@@ -1830,7 +1852,10 @@ void Translator::MatchRule(char *word[], const char *group, char *rule, MatchRec
letter_group = *rule++ - 'A';
if(IsLetter(letter_w,letter_group))
{
match.points += (20-distance_left);
lg_pts = 20;
if(letter_group==2)
lg_pts = 19; // fewer points for C, general consonant
match.points += (lg_pts-distance_left);
pre_ptr -= letter_xbytes;
}
else
+3 -1
View File
@@ -267,7 +267,9 @@ static TONE_TABLE tone_table[N_TONE_TABLE] = {
{PITCHfrise, 38,10, PITCHfrise2, 36,10, // comma, or question
20, 25, 34, 20, drops_0, 3, 3, 15, 25, 0},
{PITCHdrop, 38, 1, PITCHdrop, 42,25, // exclamation
// {PITCHdrop, 38, 1, PITCHdrop, 42,25, // exclamation
// 20, 25, 34, 22, drops_0, 3, 3, 12, 8, 0},
{PITCHfall, 30, 5, PITCHfall, 30, 7, // statement
20, 25, 34, 22, drops_0, 3, 3, 12, 8, 0},
+1 -1
View File
@@ -35,7 +35,7 @@
#include "translate.h"
#include "wave.h"
const char *version_string = "1.24.04 08.May.07";
const char *version_string = "1.24.06 10.May.07";
const int version_phdata = 0x012201;
int option_device_number = -1;
+50 -10
View File
@@ -241,8 +241,8 @@ Translator *SelectTranslator(const char *name)
case L('h','i'):
{
static const char dev_consonants2[] = {0x02,0x03,0x58,0x59,0x5a,0x5b,0x5c,0x5d,0x5e,0x5f};
static wchar_t replace_chars_hi[11] = {0x966,0x967,0x968,0x969,0x96a,0x96b,0x96c,0x96d,0x96e,0x96f,0}; // digits 0-9
static wchar_t replacement_chars_hi[11] = {0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0};
static const wchar_t replace_chars_hi[11] = {0x966,0x967,0x968,0x969,0x96a,0x96b,0x96c,0x96d,0x96e,0x96f,0}; // digits 0-9
static const unsigned int replacement_chars_hi[11] = {0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0};
static int stress_lengths_hi[8] = {190, 190, 210, 210, 0, 0, 230, 250};
static int stress_amps_hi[8] = {17,14, 20,20, 20,24, 24,22 };
@@ -269,8 +269,8 @@ Translator *SelectTranslator(const char *name)
{
static int stress_amps_hu[8] = {16,16, 20,20, 20,24, 24,22 };
static int stress_lengths_hu[8] = {180,150, 200,180, 0,0, 230,270};
static wchar_t replace_chars_hu[] = {0xd4,0xf4,0xdb,0xfb,0};
static wchar_t replacement_chars_hu[] = {0x150,0x151,0x170,0x171,0}; // allow o,u-circumflex for o,u-double-acute
static const wchar_t replace_chars_hu[] = {0xd4,0xf4,0xdb,0xfb,0};
static const unsigned int replacement_chars_hu[] = {0x150,0x151,0x170,0x171,0}; // allow o,u-circumflex for o,u-double-acute
tr = new Translator();
SetupTranslator(tr,stress_lengths_hu,stress_amps_hu);
@@ -290,6 +290,26 @@ Translator *SelectTranslator(const char *name)
}
break;
case L('i','s'): // Icelandic
{
static int stress_amps_is[] = {16,16, 20,20, 20,24, 24,22 };
static int stress_lengths_is[8] = {180,140, 200,200, 0,0, 240,260};
static const wchar_t is_L08[] = {'c','f','h','k','p','s','t','x',0xfe,0}; // voiceless conants, including 'þ'
tr = new Translator();
SetupTranslator(tr,stress_lengths_is,stress_amps_is);
tr->langopts.stress_rule = 1;
tr->langopts.param[LOPT_IT_LENGTHEN] = 0x11; // remove lengthen indicator from unstressed vowels
SetLetterBits(tr,4,"kpst"); // Letter group F
SetLetterBits(tr,3,"jvr"); // Letter group H
tr->letter_groups[8] = is_L08;
SetLetterVowel(tr,'y');
tr->langopts.numbers = 0xe9;
tr->langopts.numbers2 = 0x2;
}
break;
case L('i','t'): // Italian
{
static int stress_lengths_it[8] = {150, 140, 180, 180, 0, 0, 270, 320};
@@ -378,8 +398,8 @@ Translator *SelectTranslator(const char *name)
{
static int stress_lengths_ro[8] = {170, 170, 180, 180, 0, 0, 240, 260};
static int stress_amps_ro[8] = {15,13, 18,18, 20,22, 22,22 };
static wchar_t replace_chars_ro[5] = {0x218,0x219,0x21a,0x21b,0};
static wchar_t replacement_chars_ro[5] = {0x15e,0x15f,0x162,0x163,0}; // replace s-comma, t-comma by s-cedilla, t-cedilla
static const wchar_t replace_chars_ro[5] = {0x218,0x219,0x21a,0x21b,0};
static const unsigned int replacement_chars_ro[5] = {0x15e,0x15f,0x162,0x163,0}; // replace s-comma, t-comma by s-cedilla, t-cedilla
tr = new Translator();
SetupTranslator(tr,stress_lengths_ro,stress_amps_ro);
@@ -403,9 +423,24 @@ Translator *SelectTranslator(const char *name)
case L('c','s'): // Czech
case L('h','r'): // Croatian
{
static const wchar_t replace_chars_hr[] =
{0x430,0x431,0x446,0x45b,0x447,0x434,0x452,0x45f,
0x435,0x444,0x433,0x445,0x438,0x458,0x43a,0x43b,
0x459,0x43c,0x43d,0x45a,0x43e,0x43f,0x440,0x441,
0x448,0x442,0x443,0x432,0x437,0x436,
0x453,0x455,0x45c,0}; // ѓ ѕ ќ
static const unsigned int replacement_chars_hr[] =
{'a','b','c',0x107,0x10d,'d',0x111,'d'+(0x17e<<16),
'e','f','g','h','i','j','k','l',
'l'+('j'<<16),'m','n','n'+('j'<<16),'o','p','r','s',
0x161,'t','u','v','z',0x17e,
0x111,'d'+('z'<<16),0x107,0};
static int stress_amps_sk[8] = {16,16, 20,20, 20,24, 24,22 };
static int stress_lengths_sk[8] = {180,180, 200,180, 0,0, 220,230};
static char *sk_voiced = "bdgjlmnrvwzaeiouy";
// static int stress_lengths_sk[8] = {180,180, 200,200, 0,0, 210,220};
static int stress_lengths_sk[8] = {190,190, 210,210, 0,0, 210,210};
static char *sk_voiced = "bdgjlmnrvwzaeiouy";
tr = new Translator();
SetupTranslator(tr,stress_lengths_sk,stress_amps_sk);
@@ -420,7 +455,12 @@ Translator *SelectTranslator(const char *name)
tr->langopts.numbers = 0x1c0d + 0x84000;
if(name2 == L('h','r'))
{
tr->langopts.numbers2 = 0xa; // variant numbers before thousands,milliards
tr->langopts.replace_chars = replace_chars_hr;
tr->langopts.replacement_chars = replacement_chars_hr;
}
if(name2 == L('c','s'))
tr->langopts.numbers2 = 0x8; // variant numbers before milliards
@@ -835,8 +875,8 @@ Translator_Afrikaans::Translator_Afrikaans() : Translator()
}
int Translator_Afrikaans::TranslateChar(char *ptr, int prev_in, int c, int next_in)
{//===============================================================================
int Translator_Afrikaans::TranslateChar(char *ptr, int prev_in, int c, int next_in, int *insert)
{//=============================================================================================
// look for 'n and replace by a special character (unicode: schwa)
if(!iswalpha(prev_in))
+1 -1
View File
@@ -51,7 +51,7 @@ class Translator_Afrikaans: public Translator
public:
Translator_Afrikaans();
private:
int TranslateChar(char *ptr, int prev_in, int c, int next_in);
int TranslateChar(char *ptr, int prev_in, int c, int next_in, int *insert);
}; // end of class Translator_Afrikaans
+30 -7
View File
@@ -393,7 +393,7 @@ Translator::Translator()
SetLetterBits(this,1,"bcdfgjklmnpqstvxz"); // B hard consonants, excluding h,r,w
SetLetterBits(this,2,"bcdfghjklmnpqrstvwxz"); // C all consonants
SetLetterBits(this,3,"hlmnr"); // H 'soft' consonants
SetLetterBits(this,4,"cfhkpqstx"); // F voiceless consonants
SetLetterBits(this,5,"bdgjlmnrvwyz"); // G voiced
SetLetterBits(this,6,"eiy"); // Letter group Y, front vowels
SetLetterBits(this,7,"aeiouy"); // vowels, including y
@@ -1470,18 +1470,28 @@ static int EmbeddedCommand(unsigned int &source_index)
int Translator::TranslateChar(char *ptr, int prev_in, int c, int next_in)
{//======================================================================
int Translator::TranslateChar(char *ptr, int prev_in, int c, int next_in, int *insert)
{//===================================================================================
// To allow language specific examination and replacement of characters
const wchar_t *p;
unsigned int new_c;
if(c == 0) return(0);
if(langopts.replace_chars != NULL)
{
// there is a list of character codes to be substituted with alternative codes
if((p = wcschr(langopts.replace_chars,c)) != NULL)
{
return(langopts.replacement_chars[p - langopts.replace_chars]);
new_c = langopts.replacement_chars[p - langopts.replace_chars];
if(new_c & 0xffe00000)
{
// there is a second character to be inserted
*insert = (new_c >> 16);
new_c &= 0xffff;
}
return(new_c);
}
}
return(c);
@@ -1500,6 +1510,7 @@ void *Translator::TranslateClause(FILE *f_text, const void *vp_input, int *tone_
int prev_out2;
int prev_in2=0;
int next_in;
int char_inserted=0;
int clause_pause;
int pre_pause=0;
int pre_pause_add=0;
@@ -1626,9 +1637,19 @@ void *Translator::TranslateClause(FILE *f_text, const void *vp_input, int *tone_
utf8_in(&prev_in,&source[source_index-1],1); // prev_in = source[source_index-1];
}
prev_source_index = source_index;
source_index += utf8_in(&cc,&source[source_index],0); // cc = source[source_index++];
c = cc;
if(char_inserted)
{
c = char_inserted;
char_inserted = 0;
}
else
{
source_index += utf8_in(&cc,&source[source_index],0); // cc = source[source_index++];
c = cc;
}
utf8_in(&next_in,&source[source_index],0);
if((c == CTRL_EMBEDDED) || (c == ctrl_embedded))
{
// start of embedded command in the text
@@ -1705,7 +1726,9 @@ if((c == '/') && (langopts.testing & 2) && isdigit(next_in) && IsAlpha(prev_out)
c = '\'';
}
c = TranslateChar(&source[source_index],prev_in,c,next_in); // optional language specific function
c = TranslateChar(&source[source_index], prev_in,c, next_in, &char_inserted); // optional language specific function
if(char_inserted)
next_in = char_inserted;
if(!IsAlpha(c) && !iswspace(c) && (c != '\''))
{
+4 -3
View File
@@ -212,6 +212,7 @@ extern const int param_defaults[N_SPEECH_PARAM];
#define N_LOPTS 12
#define LOPT_DIERESES 1
// 1=remove [:] from unstressed syllables, 2= remove from unstressed or non-penultimate syllables
// bit 4=0, if stress < 4, bit 4=1, if not the highest stress in the word
#define LOPT_IT_LENGTHEN 2
// 1=german
#define LOPT_PREFIXES 3
@@ -295,8 +296,8 @@ typedef struct {
char tone_numbers;
char ideographs; // treat as separate words
int testing; // testing options: bit 1= specify stressed syllable in the form: "outdoor/2"
const wchar_t *replace_chars; // characters to be substitutes
const wchar_t *replacement_chars; // substitutes for replace_chars
const wchar_t *replace_chars; // characters to be substitutes
const unsigned int *replacement_chars; // substitutes for replace_chars
} LANGUAGE_OPTIONS;
@@ -392,7 +393,7 @@ protected:
virtual int Unpronouncable(char *word);
virtual void SetWordStress(char *output, unsigned int dictionary_flags, int tonic, int prev_stress);
virtual int RemoveEnding(char *word, int end_type, char *word_copy);
virtual int TranslateChar(char *ptr, int prev_in, int c, int next_in);
virtual int TranslateChar(char *ptr, int prev_in, int c, int next_in, int *insert);
virtual int TranslateNumber(char *word1, char *ph_out, unsigned int *flags, int wflags);
virtual int ChangePhonemes(PHONEME_LIST2 *phlist, int n_ph, int index, PHONEME_TAB *ph, CHANGEPH *ch);
+1 -1
View File
@@ -48,7 +48,7 @@ MNEM_TAB genders [] = {
{NULL, 0 }};
//int tone_points[10] = {250,140, 1200,110, -1,0, -1,0, -1,0};
int tone_points[10] = {600,180, 1200,140, 2000,110, 3000,110, -1,0};
int tone_points[10] = {600,170, 1200,135, 2000,110, 3000,110, -1,0};
// limit the rate of change for each formant number
//static int formant_rate_22050[9] = {50, 104, 165, 230, 220, 220, 220, 220, 220}; // values for 22kHz sample rate