mjohnson/espeak-ng
1
0
Fork 0
mirror of https://github.com/rhasspy/espeak-ng.git synced 2026-11-09 07:52:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

[1.29.12] *_list files can translate words to "sounds like" words as well as phonemes. $textmode, $phonememode.

Browse Source 
Lang eo: use .replace in eo_rules for "cx" etc.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@94 d46cf337-b52f-0410-862d-fd96e6ae7743
This commit is contained in:
jonsd
2007-10-29 09:42:18 +00:00
parent 5bdf9d63ab
commit 9543dcc00d
15 changed files with 240 additions and 131 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/compiledict.cpp
+73 -17
View File
@@ -44,6 +44,7 @@ static int error_count;
static int transpose_offset; // transpose character range for LookupDictList()
static int transpose_min;
static int transpose_max;
static int text_mode = 0;
int hash_counts[N_HASH_DICT];
char *hash_chains[N_HASH_DICT];
@@ -94,9 +95,13 @@ MNEM_TAB mnem_flags[] = {
{"$pastf", 27}, /* past tense follows */
{"$verbextend",28}, /* extend influence of 'verb follows' */
{"$brk", 30}, /* a shorter $pause */
{"$text", 29}, // word translates to replcement text, not phonemes
{"$brk", 30}, // a shorter $pause
// doesn't set dictionary_flags
{"$?", 100}, // conditional rule, followed by byte giving the condition number
{"$textmode", 200},
{"$phonememode", 201},
{NULL, -1}
};
@@ -169,17 +174,37 @@ int compile_line(char *linebuf, char *dict_line, int *hash)
int len_word;
int len_phonetic;
int text_not_phonemes; // this word specifies replacement text, not phonemes
char *mnemptr;
char *comment;
unsigned char flag_codes[100];
char encoded_ph[200];
unsigned char bad_phoneme[4];
p = linebuf;
comment = NULL;
text_not_phonemes = 0;
phonetic = word = "";
p = linebuf;
// while(isspace2(*p)) p++;
#ifdef deleted
if(*p == '$')
{
if(memcmp(p,"$textmode",9) == 0)
{
text_mode = 1;
return(0);
}
if(memcmp(p,"$phonememode",12) == 0)
{
text_mode = 0;
return(0);
}
}
#endif
step = 0;
c = 0;
@@ -224,7 +249,26 @@ int compile_line(char *linebuf, char *dict_line, int *hash)
ix = LookupMnem(mnem_flags,mnemptr);
if(ix > 0)
flag_codes[n_flag_codes++] = ix;
{
if(ix == 200)
{
text_mode = 1;
}
else
if(ix == 201)
{
text_mode = 0;
}
else
if(ix == BITNUM_FLAG_DICTTEXT)
{
text_not_phonemes = 1;
}
else
{
flag_codes[n_flag_codes++] = ix;
}
}
else
{
fprintf(f_log,"%5d: Unknown keyword: %s\n",linenum,mnemptr);
@@ -325,25 +369,35 @@ int compile_line(char *linebuf, char *dict_line, int *hash)
#endif
return(0); /* blank line */
}
EncodePhonemes(phonetic,encoded_ph,bad_phoneme);
if(strchr(encoded_ph,phonSWITCH) != 0)
if(text_not_phonemes || text_mode)
{
flag_codes[n_flag_codes++] = BITNUM_FLAG_ONLY_S;
strcpy(encoded_ph,phonetic); // this is replacement text, so don't encode as phonemes
flag_codes[n_flag_codes++] = BITNUM_FLAG_DICTTEXT;
}
for(ix=0; ix<255; ix++)
else
{
c = encoded_ph[ix];
if(c == 0) break;
if(c == 255)
EncodePhonemes(phonetic,encoded_ph,bad_phoneme);
if(strchr(encoded_ph,phonSWITCH) != 0)
{
/* unrecognised phoneme, report error */
fprintf(f_log,"%5d: Bad phoneme [%c] (0x%x) in: %s %s\n",linenum,bad_phoneme[0],bad_phoneme[0],word,phonetic);
error_count++;
flag_codes[n_flag_codes++] = BITNUM_FLAG_ONLY_S; // don't match on suffixes (except 's') when switching languages
}
// check for errors in the phonemes codes
for(ix=0; ix<sizeof(encoded_ph); ix++)
{
c = encoded_ph[ix];
if(c == 0) break;
if(c == 255)
{
/* unrecognised phoneme, report error */
fprintf(f_log,"%5d: Bad phoneme [%c] (0x%x) in: %s %s\n",linenum,bad_phoneme[0],bad_phoneme[0],word,phonetic);
error_count++;
}
}
}
if((word[0] & 0x80)==0) // 7 bit ascii only
{
// If first letter is uppercase, convert to lower case. (Only if it's 7bit ascii)
@@ -504,6 +558,8 @@ int compile_dictlist_file(const char *path, const char* filename)
char buf[sizeof(path_home)+45];
char dict_line[128];
text_mode = 0;
sprintf(buf,"%s%s",path,filename);
if((f_in = fopen(buf,"r")) == NULL)
return(-1);
src/dictionary.cpp
+53 -20
View File
@@ -2795,7 +2795,11 @@ int Translator::LookupDict2(char *word, char *word2, char *phonetic, unsigned in
DecodePhonemes(phonetic,ph_decoded);
if(flags != NULL)
flags1 = *flags;
fprintf(f_trans,"Found: %s [%s] %s\n",word1,ph_decoded,print_dflags(flags1));
if((dictionary_flags & FLAG_DICTTEXT) == 0)
{
fprintf(f_trans,"Found: %s [%s] %s\n",word1,ph_decoded,print_dflags(flags1));
}
}
return(1);
@@ -2805,8 +2809,8 @@ int Translator::LookupDict2(char *word, char *word2, char *phonetic, unsigned in
int Translator::LookupDictList(char *word1, char *ph_out, unsigned int *flags, int end_flags)
//===========================================================================================
int Translator::LookupDictList(char **wordptr, char *ph_out, unsigned int *flags, int end_flags)
//==============================================================================================
/* Lookup a specified word in the word dictionary.
Returns phonetic data in 'phonetic' and bits in 'flags'
@@ -2815,14 +2819,16 @@ int Translator::LookupDictList(char *word1, char *ph_out, unsigned int *flags, i
{
int length;
int found;
char *word1;
char *word2;
unsigned char c;
int nbytes;
int c2;
char word[N_WORD_BYTES];
static char word_replacement[N_WORD_BYTES];
length = 0;
word2 = word1;
word2 = word1 = *wordptr;
while((word2[nbytes = utf8_in(&c2,word2,0)]==' ') && (word2[nbytes+1]=='.'))
{
@@ -2857,25 +2863,52 @@ int Translator::LookupDictList(char *word1, char *ph_out, unsigned int *flags, i
found = LookupDict2(word,word1,ph_out,flags,end_flags);
if(found) return(1);
ph_out[0] = 0;
// try modifications to find a recognised word
if((end_flags & FLAG_SUFX_E_ADDED) && (word[length-1] == 'e'))
if(found == 0)
{
// try removing an 'e' which has been added by RemoveEnding
word[length-1] = 0;
found = LookupDict2(word,word1,ph_out,flags,end_flags);
if(found) return(1);
ph_out[0] = 0;
// try modifications to find a recognised word
if((end_flags & FLAG_SUFX_E_ADDED) && (word[length-1] == 'e'))
{
// try removing an 'e' which has been added by RemoveEnding
word[length-1] = 0;
found = LookupDict2(word,word1,ph_out,flags,end_flags);
}
else
if((end_flags & SUFX_D) && (word[length-1] == word[length-2]))
{
// try removing a double letter
word[length-1] = 0;
found = LookupDict2(word,word1,ph_out,flags,end_flags);
}
}
if((end_flags & SUFX_D) && (word[length-1] == word[length-2]))
if(found)
{
// try removing a double letter
word[length-1] = 0;
found = LookupDict2(word,word1,ph_out,flags,end_flags);
if(found) return(1);
if(*flags & FLAG_DICTTEXT)
{
// the word translates to replacement text, not to phonemes
if(end_flags & FLAG_ALLOW_DICTTEXT)
{
// only use replacement text if this is the original word, not if a prefix or suffix has been removed
word_replacement[0] = 0;
word_replacement[1] = ' ';
strcpy(&word_replacement[2],ph_out); // replacement word, preceded by zerochar and space
*wordptr = &word_replacement[2];
if(option_phonemes == 2)
{
fprintf(f_trans,"Replace: %s %s\n",word,*wordptr);
}
}
ph_out[0] = 0;
return(0);
}
return(1);
}
ph_out[0] = 0;
@@ -2887,7 +2920,7 @@ int Translator::LookupDictList(char *word1, char *ph_out, unsigned int *flags, i
int Translator::Lookup(char *word, char *ph_out)
{//=============================================
unsigned int flags;
return(LookupDictList(word,ph_out,&flags,0));
return(LookupDictList(&word,ph_out,&flags,0));
}
src/readclause.cpp
+1 -1
View File
@@ -470,7 +470,7 @@ const char *Translator::LookupSpecial(char *string)
char phonemes2[55];
static char buf[60];
if(LookupDictList(string,phonemes,&flags,0))
if(LookupDictList(&string,phonemes,&flags,0))
{
SetWordStress(phonemes,flags,-1,0);
DecodePhonemes(phonemes,phonemes2);
src/synthdata.cpp
+1 -1
View File
@@ -35,7 +35,7 @@
#include "translate.h"
#include "wave.h"
const char *version_string = "1.29.11 23.Oct.07";
const char *version_string = "1.29.12 29.Oct.07";
const int version_phdata = 0x012901;
int option_device_number = -1;
src/tr_languages.cpp
+1 -1
View File
@@ -219,7 +219,7 @@ Translator *SelectTranslator(const char *name)
tr->langopts.unstressed_wd1 = 1;
tr->langopts.unstressed_wd2 = 2;
tr->langopts.numbers = 0x1409;
tr->langopts.numbers = 0x1c09 + NUM_ROMAN;
}
break;
src/translate.cpp
+46 -46
View File
@@ -592,7 +592,7 @@ int Translator::TranslateWord(char *word1, int next_pause, WORD_TAB *wtab)
unsigned int dictionary_flags2=0;
int end_type=0;
int prefix_type=0;
char *word;
char *wordx;
char phonemes[N_WORD_PHONEMES];
char *ph_limit;
char *phonemes_ptr;
@@ -618,20 +618,19 @@ int Translator::TranslateWord(char *word1, int next_pause, WORD_TAB *wtab)
static char word_iz[4] = {0,'i','z',0};
static char word_ss[4] = {0,'s','s',0};
word = word1;
prefix_phonemes[0] = 0;
end_phonemes[0] = 0;
ph_limit = &phonemes[N_WORD_PHONEMES];
// count the length of the word
utf8_in(&first_char,word,0);
wordx = word1;
utf8_in(&first_char,wordx,0);
word_length = 0;
while((*word != 0) && (*word != ' '))
while((*wordx != 0) && (*wordx != ' '))
{
word += utf8_in(&last_char,word,0);
wordx += utf8_in(&last_char,wordx,0);
word_length++;
}
word = word1;
// try an initial lookup in the dictionary list, we may find a pronunciation specified, or
// we may just find some flags
@@ -643,7 +642,8 @@ int Translator::TranslateWord(char *word1, int next_pause, WORD_TAB *wtab)
else
{
spell_word = 0;
found = LookupDictList(word,phonemes,&dictionary_flags,wflags << 16);
found = LookupDictList(&word1,phonemes,&dictionary_flags,FLAG_ALLOW_DICTTEXT | wflags << 16); // the original word
if(phonemes[0] == phonSWITCH)
{
// change to another language in order to translate this word
@@ -669,7 +669,7 @@ if((wmark > 0) && (wmark < 8))
if(word_phonemes[0] == phonSWITCH)
return(0);
found = TranslateNumber(word,phonemes,&dictionary_flags,wflags);
found = TranslateNumber(word1,phonemes,&dictionary_flags,wflags);
}
if(!found & ((word_flags & FLAG_UPPERS) != FLAG_FIRST_UPPER))
@@ -678,7 +678,7 @@ if((wmark > 0) && (wmark < 8))
if((langopts.numbers & NUM_ROMAN) || ((langopts.numbers & NUM_ROMAN_UC) && (word_flags & FLAG_ALL_UPPER)))
{
if((found = TranslateRoman(word,phonemes)) != 0)
if((found = TranslateRoman(word1,phonemes)) != 0)
dictionary_flags |= FLAG_ABBREV; // don't spell capital Roman numbers as individual letters
}
}
@@ -694,14 +694,14 @@ if((wmark > 0) && (wmark < 8))
if(spell_word > 0)
{
// Speak as individual letters
word = word1;
wordx = word1;
posn = 0;
phonemes[0] = 0;
end_type = 0;
while(*word != ' ')
while(*wordx != ' ')
{
word += TranslateLetter(word, phonemes,spell_word);
wordx += TranslateLetter(wordx, phonemes,spell_word);
if(phonemes[0] == phonSWITCH)
{
// change to another language in order to translate this word
@@ -719,12 +719,14 @@ if((wmark > 0) && (wmark < 8))
posn = 0;
length = 999;
while(((length < 3) && (length > 0))|| (word_length > 1 && Unpronouncable(word)))
wordx = word1;
while(((length < 3) && (length > 0))|| (word_length > 1 && Unpronouncable(wordx)))
{
char *p;
// This word looks "unpronouncable", so speak letters individually until we
// find a remainder that we can pronounce.
word += TranslateLetter(word,phonemes,0);
wordx += TranslateLetter(wordx,phonemes,0);
if(phonemes[0] == phonSWITCH)
{
// change to another language in order to translate this word
@@ -732,7 +734,7 @@ if((wmark > 0) && (wmark < 8))
return(0);
}
p = &word[word_length-3];
p = &wordx[word_length-3];
if(memcmp(p,"'s ",3) == 0)
{
// remove a 's suffix and pronounce this separately (not as an individual letter)
@@ -743,18 +745,18 @@ if((wmark > 0) && (wmark < 8))
}
length=0;
while(word[length] != ' ') length++;
while(wordx[length] != ' ') length++;
if(length > 0)
word[-1] = ' '; // prevent this affecting the pronunciation of the pronuncable part
wordx[-1] = ' '; // prevent this affecting the pronunciation of the pronuncable part
}
SetSpellingStress(phonemes,0);
// anything left ?
if(*word != ' ')
if(*wordx != ' ')
{
// Translate the stem
unpron_length = strlen(phonemes);
end_type = TranslateRules(word, phonemes, N_WORD_PHONEMES, end_phonemes, wflags, dictionary_flags);
end_type = TranslateRules(wordx, phonemes, N_WORD_PHONEMES, end_phonemes, wflags, dictionary_flags);
if(phonemes[0] == phonSWITCH)
{
@@ -763,7 +765,7 @@ if((wmark > 0) && (wmark < 8))
return(0);
}
c_temp = word[-1];
c_temp = wordx[-1];
found = 0;
confirm_prefix = 1;
@@ -779,12 +781,12 @@ if((wmark > 0) && (wmark < 8))
// remove any standard suffix and confirm that the prefix is still recognised
phonemes2[0] = 0;
end2 = TranslateRules(word, phonemes2, N_WORD_PHONEMES, end_phonemes2, wflags|FLAG_NO_PREFIX|FLAG_NO_TRACE, dictionary_flags);
end2 = TranslateRules(wordx, phonemes2, N_WORD_PHONEMES, end_phonemes2, wflags|FLAG_NO_PREFIX|FLAG_NO_TRACE, dictionary_flags);
if(end2)
{
RemoveEnding(word,end2,word_copy);
end_type = TranslateRules(word, phonemes, N_WORD_PHONEMES, end_phonemes, wflags|FLAG_NO_TRACE, dictionary_flags);
memcpy(word,word_copy,strlen(word_copy));
RemoveEnding(wordx,end2,word_copy);
end_type = TranslateRules(wordx, phonemes, N_WORD_PHONEMES, end_phonemes, wflags|FLAG_NO_TRACE, dictionary_flags);
memcpy(wordx,word_copy,strlen(word_copy));
if((end_type & SUFX_P) == 0)
{
// after removing the suffix, the prefix is no longer recognised.
@@ -812,30 +814,30 @@ if((wmark > 0) && (wmark < 8))
expect_verb = 1; // use the verb form of the word
}
word[-1] = c_temp;
wordx[-1] = c_temp;
for(ix=(prefix_type & 0xf); ix>0; ix--) // num. of characters to remove
{
word++;
while((*word & 0xc0) == 0x80) word++; // for multibyte characters
wordx++;
while((*wordx & 0xc0) == 0x80) wordx++; // for multibyte characters
}
c_temp = word[-1];
word[-1] = ' ';
c_temp = wordx[-1];
wordx[-1] = ' ';
confirm_prefix = 1;
end_type = 0;
found = LookupDictList(word,phonemes,&dictionary_flags2,SUFX_P | (wflags << 16));
found = LookupDictList(&wordx,phonemes,&dictionary_flags2,SUFX_P | (wflags << 16)); // without prefix
if(dictionary_flags==0)
dictionary_flags = dictionary_flags2;
else
prefix_flags = 1;
if(found == 0)
{
end_type = TranslateRules(word, phonemes, N_WORD_PHONEMES, end_phonemes, 0, dictionary_flags);
end_type = TranslateRules(wordx, phonemes, N_WORD_PHONEMES, end_phonemes, 0, dictionary_flags);
if(phonemes[0] == phonSWITCH)
{
// change to another language in order to translate this word
word[-1] = c_temp;
wordx[-1] = c_temp;
strcpy(word_phonemes,phonemes);
return(0);
}
@@ -848,7 +850,7 @@ char phonemes2[N_WORD_PHONEMES];
strcpy(phonemes2,phonemes);
// The word has a standard ending, re-translate without this ending
end_flags = RemoveEnding(word,end_type,word_copy);
end_flags = RemoveEnding(wordx,end_type,word_copy);
phonemes_ptr = &phonemes[unpron_length];
phonemes_ptr[0] = 0;
@@ -856,26 +858,24 @@ strcpy(phonemes2,phonemes);
if(prefix_phonemes[0] != 0)
{
// lookup the stem without the prefix removed
word[-1] = c_temp;
found = LookupDictList(word1,phonemes_ptr,&dictionary_flags2,end_flags | (wflags << 16));
word[-1] = ' ';
wordx[-1] = c_temp;
found = LookupDictList(&word1,phonemes_ptr,&dictionary_flags2,end_flags | (wflags << 16)); // include prefix, but not suffix
wordx[-1] = ' ';
if(dictionary_flags==0)
dictionary_flags = dictionary_flags2;
if(found)
prefix_phonemes[0] = 0; // matched whole word, don't need prefix now
// if(found || (dictionary_flags2 != 0))
// prefix_flags = 1; // ?? this looks wrong
if((found==0) && (dictionary_flags2 != 0))
prefix_flags = 1;
}
if(found == 0)
{
found = LookupDictList(word,phonemes_ptr,&dictionary_flags2,end_flags | (wflags << 16));
found = LookupDictList(&wordx,phonemes_ptr,&dictionary_flags2,end_flags | (wflags << 16)); // without prefix and suffix
if(phonemes_ptr[0] == phonSWITCH)
{
// change to another language in order to translate this word
memcpy(word,word_copy,strlen(word_copy));
memcpy(wordx,word_copy,strlen(word_copy));
strcpy(word_phonemes,phonemes_ptr);
return(0);
}
@@ -895,16 +895,16 @@ strcpy(phonemes2,phonemes);
else
{
if(end_flags & FLAG_SUFX)
TranslateRules(word, phonemes, N_WORD_PHONEMES, NULL,wflags | FLAG_SUFFIX_REMOVED, dictionary_flags);
TranslateRules(wordx, phonemes, N_WORD_PHONEMES, NULL,wflags | FLAG_SUFFIX_REMOVED, dictionary_flags);
else
TranslateRules(word, phonemes, N_WORD_PHONEMES, NULL,wflags,dictionary_flags);
TranslateRules(wordx, phonemes, N_WORD_PHONEMES, NULL,wflags,dictionary_flags);
if(phonemes[0] == phonSWITCH)
{
// change to another language in order to translate this word
strcpy(word_phonemes,phonemes);
memcpy(word,word_copy,strlen(word_copy));
word[-1] = c_temp;
memcpy(wordx,word_copy,strlen(word_copy));
wordx[-1] = c_temp;
return(0);
}
}
@@ -917,7 +917,7 @@ strcpy(phonemes2,phonemes);
end_phonemes[0] = 0;
}
}
word[-1] = c_temp;
wordx[-1] = c_temp;
}
}
@@ -1047,7 +1047,7 @@ strcpy(phonemes2,phonemes);
expect_past = 0;
}
if((word[0] != 0) && (!(dictionary_flags & FLAG_VERB_EXT)))
if((wordx[0] != 0) && (!(dictionary_flags & FLAG_VERB_EXT)))
{
if(expect_verb > 0)
expect_verb -= 1;
src/translate.h
+8 -2
View File
@@ -36,10 +36,12 @@
// bits 0-3 stressed syllable, 7=unstressed
#define FLAG_SKIPWORDS 0x80
#define FLAG_PREPAUSE 0x100
#define FLAG_ONLY 0x200
#define FLAG_ONLY_S 0x400
#define BITNUM_FLAG_ONLY 9 // bit 9 is set
#define BITNUM_FLAG_ONLY_S 10 // bit 10 is set
#define FLAG_ONLY_S 0x400
#define FLAG_STRESS_END 0x800 /* full stress if at end of clause */
#define FLAG_STRESS_END2 0x1000 /* full stress if at end of clause, or only followed by unstressed */
#define FLAG_UNSTRESS_END 0x2000 /* reduce stress at end of clause */
@@ -62,6 +64,9 @@
#define FLAG_PASTF 0x8000000 /* past tense follows */
#define FLAG_VERB_EXT 0x10000000 /* extend the 'verb follows' */
#define FLAG_DICTTEXT 0x20000000 // word translates to replacement text, not phonemes
#define BITNUM_FLAG_DICTTEXT 29
#define FLAG_PAUSE1 0x40000000 // shorter prepause
#define FLAG_FOUND 0x80000000 /* pronunciation was found in the dictionary list */
@@ -92,6 +97,7 @@
#define SUFX_Q 0x4000 // don't retranslate
#define SUFX_T 0x10000 // don't affect the stress position in the stem
#define FLAG_ALLOW_DICTTEXT 0x02 // allow dictionary to translate to text rather than phonemes
#define FLAG_SUFX 0x04
#define FLAG_SUFX_S 0x08
#define FLAG_SUFX_E_ADDED 0x10
@@ -417,7 +423,7 @@ protected:
virtual int ChangePhonemes(PHONEME_LIST2 *phlist, int n_ph, int index, PHONEME_TAB *ph, CHANGEPH *ch);
int IsVowel(int letter);
int LookupDictList(char *word1, char *ph_out, unsigned int *flags, int end_flags);
int LookupDictList(char **wordptr, char *ph_out, unsigned int *flags, int end_flags);
int Lookup(char *word, char *ph_out);