[1.47.03c]

Phoneme definition files: added prev2PhW() and next3PhW().
Added language: ms (Malay).


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@332 d46cf337-b52f-0410-862d-fd96e6ae7743

src/compiledata.cpp

@@ -104,9 +104,8 @@ static keywtab_t k_conditions[] = {
 	{"next2PhW",  tWHICH_PHONEME, 6},
 	{"nextVowel",tWHICH_PHONEME,  7},
 	{"prevVowel",tWHICH_PHONEME,  8},
-//	{"next2PhW", tWHICH_PHONEME,  0x800},
-
-//	{"numVowels",    tTEST,  0x000},
+	{"next3PhW", tWHICH_PHONEME,  9},
+	{"prev2PhW", tWHICH_PHONEME, 10},
 
 	{"PreVoicing",   tTEST,  0xf01},
 	{"KlattSynth",   tTEST,  0xf02},
@@ -1051,8 +1050,13 @@ static wxString CompileAllDictionaries()
 			fclose(f_in);
 		}
 
-		LoadVoice(voicename,0);
-
+		if(LoadVoice(voicename,1) == NULL)
+		{
+			wxLogError(wxString::Format(_T("Can't find voice '%s' for dictionary '%s'"), wxString(voicename, wxConvLocal).c_str(), dictstr.c_str()));
+			report = report + dictstr + _T(" No Voice, ");
+			errors ++;
+		}
+		else
 		if((err = CompileDictionary(path_dsource, dictname,log,NULL,0)) > 0)
 		{
 			report = report + dictstr + wxString::Format(_T(" %d, "),err);
@@ -2742,6 +2746,7 @@ int CompilePhoneme(int compile_phoneme)
 	int count;
 	int c;
 	char *p;
+	int vowel_length_factor = 100;  // for testing
 	char number_buf[12];
 	char ipa_buf[N_ITEM_STRING+1];
 	PHONEME_TAB phoneme_out2;
@@ -2844,6 +2849,11 @@ int CompilePhoneme(int compile_phoneme)
 
 			case i_SET_LENGTH:
 				value = NextItemMax(511);
+				if(phoneme_out->type == phVOWEL)
+				{
+					value = (value * vowel_length_factor)/100;
+				}
+
 				if(after_if == 0)
 				{
 					phoneme_out->std_length = value/2;

src/compiledict.cpp

@@ -470,12 +470,12 @@ static int compile_line(char *linebuf, char *dict_line, int *hash)
 			}
 
 			ix = 0;
-			if(isdigit(*p))
+			if(IsDigit09(*p))
 			{
 				ix += (*p-'0');
 				p++;
 			}
-			if(isdigit(*p))
+			if(IsDigit09(*p))
 			{
 				ix = ix*10 + (*p-'0');
 				p++;
@@ -542,7 +542,7 @@ static int compile_line(char *linebuf, char *dict_line, int *hash)
 		case 1:
 			if((c == '-') && multiple_words)
 			{
-				if(isdigit(word[0]))
+				if(IsDigit09(word[0]))
 				{
 					multiple_numeric_hyphen = 1;
 				}
@@ -1156,7 +1156,7 @@ static void copy_rule_string(char *string, int &state)
 							sxflags |= SUFX_M;
 							break;
 						default:
-							if(isdigit(c))
+							if(IsDigit09(c))
 								value = (value*10) + (c - '0');
 							break;
 						}
@@ -1575,7 +1575,7 @@ static int compile_lettergroup(char *input, FILE *f_out)
 	char item_length[N_LETTERGP_ITEMS];
 
 	p = input;
-	if(!isdigit(p[0]) || !isdigit(p[1]))
+	if(!IsDigit09(p[0]) || !IsDigit09(p[1]))
 	{
 		fprintf(f_log,"%5d: Expected 2 digits after '.L'\n",linenum);
 		error_count++;

src/dictionary.cpp

@@ -607,7 +607,7 @@ char *WritePhMnemonic(char *phon_out, PHONEME_TAB *ph, PHONEME_LIST *plist, int
 				break;   // # is subscript-h, but only for consonants
 
 			// ignore digits after the first character
-			if(!first && isdigit(c))
+			if(!first && IsDigit09(c))
 				continue;
 
 			if((c >= 0x20) && (c < 128))
@@ -3362,7 +3362,7 @@ int LookupDictList(Translator *tr, char **wordptr, char *ph_out, unsigned int *f
 		if(((c = *word1++)==0) || (c == ' '))
 			break;
 
-		if((c=='.') && (length > 0) && (isdigit(word[length-1])))
+		if((c=='.') && (length > 0) && (IsDigit09(word[length-1])))
 			break;  // needed for lang=hu, eg. "december 2.-ig"
 
 		word[length] = c;

src/numbers.cpp

@@ -604,7 +604,7 @@ int TranslateLetter(Translator *tr, char *word, char *phonemes, int control)
 	char ph_buf[80];
 	char ph_buf2[80];
 	char ph_alphabet[80];
-	char hexbuf[6];
+	char hexbuf[12];
 	static char pause_string[] = {phonPAUSE, 0};
 
 	ph_buf[0] = 0;
@@ -981,7 +981,7 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out, WORD_TAB *wtab)
 	flags[0] = 0;
 	flags[1] = 0;
 
-	if(((tr->langopts.numbers & NUM_ROMAN_CAPITALS) && !(wtab[0].flags & FLAG_ALL_UPPER)) || isdigit(word[-2]))
+	if(((tr->langopts.numbers & NUM_ROMAN_CAPITALS) && !(wtab[0].flags & FLAG_ALL_UPPER)) || IsDigit09(word[-2]))
 		return(0);    // not '2xx'
 
 	word_start = word;
@@ -1020,7 +1020,7 @@ int TranslateRoman(Translator *tr, char *word, char *ph_out, WORD_TAB *wtab)
 		n_digits++;
 	}
 
-	if(isdigit(word[0]))
+	if(IsDigit09(word[0]))
 		return(0);      // eg. 'xx2'
 
 	acc += prev;
@@ -1756,12 +1756,12 @@ bool CheckThousandsGroup(char *word, int group_len)
 // Is this a group of 3 digits which looks like a thousands group?
 	int ix;
 
-	if(isdigit(word[group_len]) || isdigit(-1))
+	if(IsDigit09(word[group_len]) || IsDigit09(-1))
 		return(false);
 
 	for(ix=0; ix < group_len; ix++)
 	{
-		if(!isdigit(word[ix]))
+		if(!IsDigit09(word[ix]))
 			return(false);
 	}
 	return(true);
@@ -1812,7 +1812,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 	digit_lookup = buf_digit_lookup;
 	number_control = control;
 
-	for(ix=0; isdigit(word[ix]); ix++) ;
+	for(ix=0; IsDigit09(word[ix]); ix++) ;
 	n_digits = ix;
 	value = this_value = atoi(word);
 
@@ -1821,14 +1821,14 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 		group_len = 4;
 
 	// is there a previous thousands part (as a previous "word") ?
-	if((n_digits == group_len) && (word[-2] == tr->langopts.thousands_sep) && isdigit(word[-3]))
+	if((n_digits == group_len) && (word[-2] == tr->langopts.thousands_sep) && IsDigit09(word[-3]))
 	{
 		prev_thousands = 1;
 	}
 	else if((tr->langopts.thousands_sep == ' ') || (tr->langopts.numbers & NUM_ALLOW_SPACE))
 	{
 		// thousands groups can be separated by spaces
-		if((n_digits == 3) && !(wtab->flags & FLAG_MULTIPLE_SPACES) && isdigit(word[-2]))
+		if((n_digits == 3) && !(wtab->flags & FLAG_MULTIPLE_SPACES) && IsDigit09(word[-2]))
 		{
 			prev_thousands = 1;
 		}
@@ -1850,7 +1850,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 		}
 	}
 
-	if((word[ix] == '.') && !isdigit(word[ix+1]) && !isdigit(word[ix+2]) && !(wtab[1].flags & FLAG_NOSPACE))
+	if((word[ix] == '.') && !IsDigit09(word[ix+1]) && !IsDigit09(word[ix+2]) && !(wtab[1].flags & FLAG_NOSPACE))
 	{
 		// remove dot unless followed by another number
 		word[ix] = 0;
@@ -1879,7 +1879,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 			{
 				ordinal = 2;
 			}
-			else if(!isdigit(suffix[0]))  // not _#9 (tab)
+			else if(!IsDigit09(suffix[0]))  // not _#9 (tab)
 			{
 				sprintf(string,"_#%s",suffix);
 				if(Lookup(tr, string, ph_ordinal2))
@@ -1904,7 +1904,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 
 	if((word[0] == '0') && (prev_thousands == 0) && (word[1] != ' ') && (word[1] != tr->langopts.decimal_sep))
 	{
-		if((n_digits == 2) && (word[3] == ':') && isdigit(word[5]) && isspace(word[7]))
+		if((n_digits == 2) && (word[3] == ':') && IsDigit09(word[5]) && isspace(word[7]))
 		{
 			// looks like a time 02:30, omit the leading zero
 		}
@@ -1973,7 +1973,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 		}
 	}
 
-	if((word[n_digits] == tr->langopts.decimal_sep) && isdigit(word[n_digits+1]))
+	if((word[n_digits] == tr->langopts.decimal_sep) && IsDigit09(word[n_digits+1]))
 	{
 		// this "word" ends with a decimal point
 		Lookup(tr, "_dpt", ph_append);
@@ -2018,8 +2018,8 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 		char *p2;
 		// look for combinations of the number with the next word
 		p = word;
-		while(isdigit(p[1])) p++;  // just use the last digit
-		if(isdigit(p[-1]))
+		while(IsDigit09(p[1])) p++;  // just use the last digit
+		if(IsDigit09(p[-1]))
 		{
 			p2 = p - 1;
 			if(LookupDictList(tr, &p2, buf_digit_lookup, flags, FLAG_SUFX, wtab))  // lookup 2 digits
@@ -2082,7 +2082,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 		n_digits++;
 
 		decimal_count = 0;
-		while(isdigit(word[n_digits+decimal_count]))
+		while(IsDigit09(word[n_digits+decimal_count]))
 			decimal_count++;
 
 //		if(decimal_count > 1)
@@ -2101,7 +2101,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 					decimal_count--;
 					n_digits++;
 				}
-				if((decimal_count <= max_decimal_count) && isdigit(word[n_digits]))
+				if((decimal_count <= max_decimal_count) && IsDigit09(word[n_digits]))
 				{
 					LookupNum3(tr, atoi(&word[n_digits]), buf1, 0,0,0);
 					strcat(ph_out,buf1);
@@ -2152,7 +2152,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 			}
 		}
 
-		while(isdigit(c = word[n_digits]) && (strlen(ph_out) < (N_WORD_PHONEMES - 10)))
+		while(IsDigit09(c = word[n_digits]) && (strlen(ph_out) < (N_WORD_PHONEMES - 10)))
 		{
 			// speak any remaining decimal fraction digits individually
 			value = word[n_digits++] - '0';
@@ -2165,7 +2165,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 		if(Lookup(tr, "_dpt2", buf1))
 			strcat(ph_out,buf1);
 
-		if((c == tr->langopts.decimal_sep) && isdigit(word[n_digits+1]))
+		if((c == tr->langopts.decimal_sep) && IsDigit09(word[n_digits+1]))
 		{
 			Lookup(tr, "_dpt", buf1);
 			strcat(ph_out,buf1);

src/readclause.cpp

@@ -1287,10 +1287,10 @@ static int attrnumber(const wchar_t *pw, int default_value, int type)
 {//==================================================================
 	int value = 0;
 
-	if((pw == NULL) || !isdigit(*pw))
+	if((pw == NULL) || !IsDigit09(*pw))
 		return(default_value);
 
-	while(isdigit(*pw))
+	while(IsDigit09(*pw))
 	{
 		value = value*10 + *pw++ - '0';
 	}

src/synthdata.cpp

@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.47.03b  22.Mar.13";
+const char *version_string = "1.47.03c  23.Mar.13";
 const int version_phdata  = 0x014701;
 
 int option_device_number = -1;
@@ -615,6 +615,7 @@ static int CountVowelPosition(PHONEME_LIST *plist)
 static bool InterpretCondition(Translator *tr, int control, PHONEME_LIST *plist, USHORT *p_prog, WORD_PH_DATA *worddata)
 {//========================================================================================================================
 	int which;
+	int ix;
 	unsigned int data;
 	int instn;
 	int instn2;
@@ -705,6 +706,21 @@ static bool InterpretCondition(Translator *tr, int control, PHONEME_LIST *plist,
                 return(false);   // no previous vowel
 				plist = &(worddata->prev_vowel);
             break;
+
+		case 9:  // next3PhW
+			for(ix=1; ix<=3; ix++)
+			{
+				if(plist[ix].sourceix)
+					return(false);
+			}
+			plist = &plist[3];
+			break;
+
+		case 10: // prev2PhW
+			if((plist[0].sourceix) || (plist[-1].sourceix))
+				return(false);
+			plist-=2;
+			break;
         }
 
 		if((which == 0) || (which == 5))

src/tr_languages.cpp

@@ -975,6 +975,7 @@ SetLengthMods(tr,3);  // all equal
 		break;
 
 	case L('i','d'):   // Indonesian
+	case L('m','s'):   // Malay
 		{
 			static const short stress_lengths_id[8] = {160, 200,  180, 180,  0, 0,  220, 240};
 			static const unsigned char stress_amps_id[8] = {16,18, 18,18, 20,22, 22,21 };
@@ -1312,7 +1313,7 @@ SetLengthMods(tr,3);  // all equal
 			tr->langopts.stress_rule = STRESSPOSN_2R;
 			tr->langopts.stress_flags = S_FINAL_DIM_ONLY | S_FINAL_NO_2;
 			tr->langopts.length_mods0 = tr->langopts.length_mods;  // don't lengthen vowels in the last syllable
-
+			tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1;   // disable check for unpronouncable words.  Need to allow "bw'" prefix
 			tr->langopts.numbers = NUM_HUNDRED_AND | NUM_AND_UNITS | NUM_DFRACTION_2 | NUM_AND_HUNDRED;
 			tr->langopts.numbers2 = 0x200;  // say "thousands" before its number
 		}
@@ -1429,6 +1430,8 @@ SetLengthMods(tr,3);  // all equal
 			tr->langopts.vowel_pause = 1;
 			tr->langopts.stress_rule = STRESSPOSN_2R;
 			tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2;
+ 			tr->langopts.max_initial_consonants = 4; // for example: mwngi
+
 
 			tr->langopts.numbers = NUM_AND_UNITS | NUM_HUNDRED_AND | NUM_SINGLE_AND | NUM_OMIT_1_HUNDRED;
 			tr->langopts.break_numbers = 0x49249268;  // for languages which have numbers for 100,000 and 1,000,000

src/translate.cpp

@@ -374,6 +374,7 @@ void SetLengthMods(Translator *tr, int value)
 }
 
 
+
 int IsAlpha(unsigned int c)
 {//========================
 // Replacement for iswalph() which also checks for some in-word symbols
@@ -429,8 +430,8 @@ int IsAlpha(unsigned int c)
 	return(0);
 }
 
-static int IsDigit09(unsigned int c)
-{//=========================
+int IsDigit09(unsigned int c)
+{//============================
 	if((c >= '0') && (c <= '9'))
 		return(1);
 	return(0);
@@ -2311,10 +2312,10 @@ static int EmbeddedCommand(unsigned int &source_index)
 		source_index++;
 	}
 
-	if(isdigit(source[source_index]))
+	if(IsDigit09(source[source_index]))
 	{
 		value = atoi(&source[source_index]);
-		while(isdigit(source[source_index]))
+		while(IsDigit09(source[source_index]))
 			source_index++;
 	}
 
@@ -3331,7 +3332,7 @@ if((c == '/') && (tr->langopts.testing & 2) && IsDigit09(next_in) && IsAlpha(pre
 
 			while(pn < &number_buf[sizeof(number_buf)-20])
 			{
-				if(!isdigit(c = *pw++) && (c != tr->langopts.decimal_sep))
+				if(!IsDigit09(c = *pw++) && (c != tr->langopts.decimal_sep))
 					break;
 
 				*pn++ = c;

src/translate.h

@@ -709,6 +709,7 @@ void InitNamedata(void);
 void InitText(int flags);
 void InitText2(void);
 int IsDigit(unsigned int c);
+int IsDigit09(unsigned int c);
 int IsAlpha(unsigned int c);
 int IsVowel(Translator *tr, int c);
 int isspace2(unsigned int c);

src/voices.cpp

@@ -601,7 +601,7 @@ voice_t *LoadVoice(const char *vname, int control)
 
 	// which directory to look for a named voice. List of voice names, must end in a space.
 	static const char *voices_asia =
-		"fa fa-pin hi hy hy-west id ka kn ku ml ne pa ta tr vi vi-hue zh zh-yue ";
+		"fa fa-pin hi hy hy-west id ka kn ku ml ms ne pa ta tr vi vi-hue zh zh-yue ";
 	static const char *voices_europe =
 		"an bg bs ca cs cy da de el en en-us es et fi fr fr-be ga hr hu is it lt lv mk nl no pl pt-pt ro ru sk sq sr sv ";
 
@@ -1134,7 +1134,7 @@ static char *ExtractVoiceVariantName(char *vname, int variant_num, int add_dir)
 			// The voice name has a +variant suffix
 			variant_num = 0;
 			*p++ = 0;   // delete the suffix from the voice name
-			if(isdigit(*p))
+			if(IsDigit09(*p))
 			{
 				variant_num = atoi(p);  // variant number
 			}