mjohnson/espeak-ng
1
0
Fork 0
mirror of https://github.com/rhasspy/espeak-ng.git synced 2026-11-12 00:46:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

[1.42.03 release]

Browse Source 
Allow .txt release for *_rules and *_list files.
Rules lookup, consider multi-byte characters when calculating matchign points.
Fix bug in Roman Number translation.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@232 d46cf337-b52f-0410-862d-fd96e6ae7743
This commit is contained in:
jonsd
2009-12-08 17:59:06 +00:00
parent e404310b8f
commit de52d78665
19 changed files with 1362 additions and 1018 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/compiledata.cpp
+18 -3
View File
@@ -2085,6 +2085,7 @@ wxString CompileAllDictionaries()
wxString filename;
wxFileName fname;
wxString dictstr;
wxString filetype;
wxString report = _T("");
int err;
int errors = 0;
@@ -2135,18 +2136,32 @@ wxString CompileAllDictionaries()
fprintf(f_phused,"Phonemes which are used in the *_rules and *_list files\n");
}
bool cont = dir.GetFirst(&filename, _T("*_rules"), wxDIR_FILES);
bool cont = dir.GetFirst(&filename, _T("*_rules*"), wxDIR_FILES);
while ( cont )
{
fname = wxFileName(filename);
filetype = fname.GetName().AfterLast('_');
if((filetype != _T("rules")) && (filetype != _T("rules.txt")))
{
cont = dir.GetNext(&filename);
continue;
}
dictstr = fname.GetName().BeforeLast('_');
strcpy(dictname,dictstr.mb_str(wxConvLocal));
dict_count++;
strcpy(voicename,dictname);
// read the *_rules file to see if a phoneme table is specified though a voice name
sprintf(path,"%s%s_rules",path_dsource,dictname);
if((f_in = fopen(path,"r")) != NULL)
sprintf(path,"%s%s_rules.txt",path_dsource,dictname);
if((f_in = fopen(path,"r")) == NULL)
{
sprintf(path,"%s%s_rules",path_dsource,dictname);
f_in = fopen(path,"r");
}
if(f_in != NULL)
{
unsigned int ix;
unsigned int c;
src/compiledict.cpp
+20 -8
View File
@@ -690,9 +690,14 @@ static int compile_dictlist_file(const char *path, const char* filename)
text_mode = 0;
sprintf(fname,"%s%s",path,filename);
// try with and without '.txt' extension
sprintf(fname,"%s%s.txt",path,filename);
if((f_in = fopen(fname,"r")) == NULL)
return(-1);
{
sprintf(fname,"%s%s",path,filename);
if((f_in = fopen(fname,"r")) == NULL)
return(-1);
}
fprintf(f_log,"Compiling: '%s'\n",fname);
@@ -860,6 +865,9 @@ static void copy_rule_string(char *string, int &state)
case 'X':
c = RULE_NOVOWELS;
break;
case 'J':
c = RULE_SKIPCHARS;
break;
case 'L':
// expect two digits
c = *p++ - '0';
@@ -1659,14 +1667,18 @@ int CompileDictionary(const char *dsource, const char *dict_name, FILE *log, cha
if(f_log == NULL)
f_log = stderr;
// try with and without '.txt' extension
sprintf(path,"%s%s_",dsource,dict_name);
sprintf(fname_in,"%srules",path);
f_in = fopen_log(fname_in,"r");
if(f_in == NULL)
sprintf(fname_in,"%srules.txt",path);
if((f_in = fopen(fname_in,"r")) == NULL)
{
if(fname_err)
strcpy(fname_err,fname_in);
return(-1);
sprintf(fname_in,"%srules",path);
if((f_in = fopen_log(fname_in,"r")) == NULL)
{
if(fname_err)
strcpy(fname_err,fname_in);
return(-1);
}
}
sprintf(fname_out,"%s%c%s_dict",path_home,PATHSEP,dict_name);
src/dictionary.cpp
+35 -6
View File
@@ -1648,7 +1648,7 @@ static char *DecodeRule(const char *group_chars, int group_length, char *rule)
static char output[60];
static char symbols[] = {' ',' ',' ',' ',' ',' ',' ',' ',' ',
'@','&','%','+','#','S','D','Z','A','L',' ',' ',' ',' ',' ','N','K','V',' ','T','X','?','W'};
'@','&','%','+','#','S','D','Z','A','L','!',' ','?','?','J','N','K','V','?','T','X','?','W'};
static char symbols_lg[] = {'A','B','C','H','F','G','Y'};
@@ -1998,7 +1998,8 @@ static void MatchRule(Translator *tr, char *word[], int group_length, char *rule
if((letter == rb) || ((letter==(unsigned char)REPLACED_E) && (rb=='e')))
{
add_points = 21;
if((letter & 0xc0) != 0x80)
add_points = 21; // don't add point for non-initial UTF-8 bytes
consumed++;
}
else
@@ -2148,6 +2149,27 @@ static void MatchRule(Translator *tr, char *word[], int group_length, char *rule
}
break;
case RULE_SKIPCHARS:
{
// Used for lang=Tamil, used to match on the next word after an unknown word ending
// only look until the end of the word (including the end-of-word marker)
// Jx means 'skip characters until x', where 'x' may be '_' for 'end of word'
char *p = post_ptr + letter_xbytes;
char *p2 = p;
int rule_w; // skip characters until this
utf8_in(&rule_w,rule);
while((letter_w != rule_w) && (letter_w != RULE_SPACE))
{
p2 = p;
p += utf8_in(&letter_w,p);
}
if(letter_w == rule_w)
{
post_ptr = p2;
}
}
break;
case RULE_INC_SCORE:
add_points = 20; // force an increase in points
break;
@@ -2180,10 +2202,11 @@ static void MatchRule(Translator *tr, char *word[], int group_length, char *rule
default:
if(letter == rb)
{
if(letter == RULE_SPACE)
add_points = (21-distance_right);
else
if((letter & 0xc0) != 0x80)
{
// not for non-initial UTF-8 bytes
add_points = (21-distance_right);
}
}
else
failed = 1;
@@ -2348,7 +2371,13 @@ static void MatchRule(Translator *tr, char *word[], int group_length, char *rule
if(letter == RULE_SPACE)
add_points = 4;
else
add_points = (21-distance_left);
{
if((letter & 0xc0) != 0x80)
{
// not for non-initial UTF-8 bytes
add_points = (21-distance_left);
}
}
}
else
failed = 1;
src/extras.cpp
+10 -5
View File
@@ -1344,14 +1344,19 @@ void FormatDictionary(const char *dictname)
const int tab2 = 18;
const int tab3 = 28;
sprintf(fname_in,"%s/%s_rules",path_dsource,dictname);
sprintf(fname_out,"%s_formatted",fname_in);
// try with and without '.txt' extension
sprintf(fname_in,"%s/%s_rules.txt",path_dsource,dictname);
if((f_in = fopen(fname_in,"r")) == NULL)
{
wxLogError(_T("Can't open rules file: ") + wxString(fname_in,wxConvLocal));
return;
sprintf(fname_in,"%s/%s_rules",path_dsource,dictname);
if((f_in = fopen(fname_in,"r")) == NULL)
{
wxLogError(_T("Can't open rules file: ") + wxString(fname_in,wxConvLocal));
return;
}
}
sprintf(fname_out,"%s/%s_rules_formatted",path_dsource,dictname);
if((f_out = fopen(fname_out,"w")) == NULL)
{
wxLogError(_T("Can't write to file: ") + wxString(fname_out,wxConvLocal));
src/numbers.cpp
+2
View File
@@ -813,6 +813,8 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out, WORD_TAB *wtab)
prev = 0;
subtract = 0x7fff;
ph_out[0] = 0;
flags[0] = 0;
flags[1] = 0;
if((tr->langopts.numbers & NUM_ROMAN_CAPITALS) && !(wtab[0].flags & FLAG_ALL_UPPER))
return(0);
src/readclause.cpp
+9 -5
View File
@@ -92,6 +92,9 @@ static const unsigned short punct_chars[] = {',','.','?','!',':',';',
0x055e, // Armenian question
0x055b, // Armenian emphasis mark
0x0b1b, // Arabic ;
0x061f, // Arabic ?
0x1362, // Ethiopic period
0x1363,
0x1364,
@@ -131,6 +134,9 @@ static const unsigned int punct_attributes [] = { 0,
CLAUSE_QUESTION + PUNCT_IN_WORD, // Armenian question
CLAUSE_PERIOD + PUNCT_IN_WORD, // Armenian emphasis mark
CLAUSE_SEMICOLON, // Arabic ;
CLAUSE_QUESTION, // Arabic question mark
CLAUSE_PERIOD, // Ethiopic period
CLAUSE_COMMA, // Ethiopic comma
CLAUSE_SEMICOLON, // Ethiopic semicolon
@@ -570,6 +576,7 @@ static const char *LookupSpecial(Translator *tr, const char *string, char* text_
char phonemes2[55];
char *string1 = (char *)string;
flags[0] = flags[1] = 0;
if(LookupDictList(tr,&string1,phonemes,flags,0,NULL))
{
SetWordStress(tr, phonemes, flags, -1, 0);
@@ -2406,14 +2413,11 @@ if(option_ssml) parag=1;
// Because of an xml tag, we are waiting for the
// next non-blank character to decide whether to end the clause
// i.e. is dot followed by an upper-case letter?
if(c1 == '\n')
{
// end_clause_after_tag &= ~CLAUSE_DOT;
}
if(!iswspace(c1))
{
if(iswdigit(c1) || (IsAlpha(c1) && !iswlower(c1)))
if(!IsAlpha(c1) || !iswlower(c1))
// if(iswdigit(c1) || (IsAlpha(c1) && !iswlower(c1)))
{
UngetC(c2);
ungot_char2 = c1;
src/synthdata.cpp
+1 -1
View File
@@ -35,7 +35,7 @@
#include "translate.h"
#include "wave.h"
const char *version_string = "1.42.02 30.Nov.09";
const char *version_string = "1.42.03 08.Dec.09";
const int version_phdata = 0x014200;
int option_device_number = -1;
src/tr_languages.cpp
+8 -2
View File
@@ -279,7 +279,6 @@ Translator *SelectTranslator(const char *name)
break;
case L('a','r'): // Arabic
case L('u','r'): // Urdu
tr->letter_bits_offset = OFFSET_ARABIC;
tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1; // disable check for unpronouncable words
break;
@@ -538,7 +537,7 @@ Translator *SelectTranslator(const char *name)
tr->langopts.stress_rule = 6; // stress on last heaviest syllable, excluding final syllable
tr->langopts.stress_flags = 0x10004; // use 'diminished' for unstressed final syllable
tr->langopts.numbers = NUM_SWAP_TENS;
tr->langopts.break_numbers = 0x24924aa8; // for languages which have numbers for 100,000 and 100,00,000, eg Hindi
tr->langopts.break_numbers = 0x14aa8; // for languages which have numbers for 100,000 and 100,00,000, eg Hindi
tr->letter_bits_offset = OFFSET_DEVANAGARI;
if(name2 == L('p','a'))
@@ -1034,6 +1033,13 @@ SetLengthMods(tr,3); // all equal
}
break;
case L('u','r'): // Urdu
tr->letter_bits_offset = OFFSET_ARABIC;
tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1; // disable check for unpronouncable words
tr->langopts.numbers = NUM_SWAP_TENS;
tr->langopts.break_numbers = 0x52a8; // for languages which have numbers for 100,000 and 100,00,000, eg Hindi
break;
case L('v','i'): // Vietnamese
{
static const short stress_lengths_vi[8] = {150, 150, 180, 180, 210, 230, 230, 240};
src/translate.h
+8 -7
View File
@@ -135,18 +135,19 @@
#define RULE_LINENUM 8 // next 2 bytes give a line number, for debugging purposes
#define RULE_SPACE 32 // ascii space
#define RULE_SYLLABLE 9
#define RULE_STRESSED 10
#define RULE_DOUBLE 11
#define RULE_INC_SCORE 12
#define RULE_DEL_FWD 13
#define RULE_ENDING 14
#define RULE_SYLLABLE 9 // @
#define RULE_STRESSED 10 // &
#define RULE_DOUBLE 11 // %
#define RULE_INC_SCORE 12 // +
#define RULE_DEL_FWD 13 // #
#define RULE_ENDING 14 // S
#define RULE_DIGIT 15 // D digit
#define RULE_NONALPHA 16 // Z non-alpha
#define RULE_LETTERGP 17 // A B C H F G Y letter group number
#define RULE_LETTERGP2 18 // L + letter group number
#define RULE_CAPITAL 19 // word starts with a capital letter
#define RULE_CAPITAL 19 // ! word starts with a capital letter
#define RULE_REPLACEMENTS 20 // section for character replacements
#define RULE_SKIPCHARS 23 // J
#define RULE_NO_SUFFIX 24 // N
#define RULE_NOTVOWEL 25 // K
#define RULE_IFVERB 26 // V