Merge branch 'master' into build

src/Makefile

@@ -1,10 +1,11 @@
 # remove -fno-exceptions  if it causes problems with stdio library on some platforms (ARM)
 
+DATADIR=/usr/share/espeak-data
+
 PREFIX=/usr
 BINDIR=$(PREFIX)/bin
 INCDIR=$(PREFIX)/include/espeak
 LIBDIR=$(PREFIX)/lib
-DATADIR=$(PREFIX)/share/espeak-data
 
 RELEASE = 1.44
 BIN_NAME = speak
@@ -88,7 +89,7 @@ $(BIN2_NAME): $(OBJS3) $(LIB_NAME)
 
 
 x_%.o: %.cpp
-	$(CXX) $(CXXFLAGS) $(USE_AUDIO) -Wall -fpic -fvisibility=hidden -pedantic \
+	$(CXX) $(CXXFLAGS) $(USE_AUDIO) -D PATH_ESPEAK_DATA=\"$(DATADIR)\" -Wall -fpic -fvisibility=hidden -pedantic \
 	-I. -D LIBRARY -c -fno-exceptions  $<  -o x_$*.o
 
 $(LIB_NAME): $(OBJS2)

src/compiledata.cpp

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2005 to 2010 by Jonathan Duddington                     *
+ *   Copyright (C) 2005 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -312,6 +312,7 @@ static keywtab_t keywords[] = {
 	{"nonsyllabic",tPHONEME_FLAG, phNONSYLLABIC},
 	{"lengthenstop",tPHONEME_FLAG, phLENGTHENSTOP},
 	{"nopause",    tPHONEME_FLAG, phNOPAUSE},
+	{"prevoice",   tPHONEME_FLAG, phPREVOICE},
 
 	{"flag1",      tPHONEME_FLAG, phFLAG1},
 	{"flag2",      tPHONEME_FLAG, phFLAG2},
@@ -375,8 +376,10 @@ int n_procs;
 int  proc_addr[N_PROCS];
 char proc_names[40][N_PROCS];
 
+#define MAX_PROG_BUF 2000
 USHORT *prog_out;
-USHORT prog_buf[2000];
+USHORT *prog_out_max;
+USHORT prog_buf[MAX_PROG_BUF+20];
 
 
 static const char *KeyToMnem(keywtab_t *ktab, int type, int value)
@@ -1064,6 +1067,8 @@ static wxString CompileAllDictionaries()
 
 static void error(const char *format, const char *string)
 {//======================================================
+	if(string==NULL)
+		string = "";
 	fprintf(f_errors,"%4d:  ",linenum-1);
 	fprintf(f_errors,format,string);
 	fprintf(f_errors,"\n");
@@ -2259,6 +2264,12 @@ int CompileIf(int elif)
 
 	while(!finish)
 	{
+		if(prog_out >= prog_out_max)
+		{
+			error("Phoneme program too large", NULL);
+			return(0);
+		}
+
 		if((key = NextItem(tCONDITION)) < 0)
 			error("Expected a condition, not '%s'",item_string);
 	
@@ -2644,6 +2655,7 @@ int CompilePhoneme(int compile_phoneme)
 	PHONEME_PROG_LOG phoneme_prog_log;
 
 	prog_out = prog_buf;
+	prog_out_max = &prog_buf[MAX_PROG_BUF-1];
 	if_level = 0;
 	if_stack[0].returned = 0;
 	after_if = 0;

src/dictionary.cpp

@@ -2740,6 +2740,8 @@ void ApplySpecialAttribute(Translator *tr, char *phonemes, int dict_flags)
 
 	switch(tr->translator_name)
 	{
+#ifdef deleted
+// this is now done in de_rules
 	case L('d','e'):
 		if(p_end[0] == PhonemeCode2('i',':'))
 		{
@@ -2750,6 +2752,7 @@ void ApplySpecialAttribute(Translator *tr, char *phonemes, int dict_flags)
 			p_end[2] = 0;
 		}
 		break;
+#endif
 
 	case L('r','o'):
 		if(p_end[0] == PhonemeCode('j'))
@@ -3249,6 +3252,10 @@ 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])))
+			break;  // needed for lang=hu, eg. "december 2.-ig"
+
 		word[length] = c;
 	}
 	word[length] = 0;

src/espeak.cpp

@@ -426,6 +426,9 @@ int main (int argc, char **argv)
 
 		if(c == '-')
 		{
+			if(p[0] == 0)
+				break;   // -- means don't interpret further - as commands
+
 			opt_string="";
 			for(ix=0; ;ix++)
 			{
@@ -478,9 +481,15 @@ int main (int argc, char **argv)
 			break;
 
 		case 'h':
-			printf("\n");
-			printf("eSpeak text-to-speech: %s\n%s",espeak_Info(NULL),help_text);
-			exit(0);
+			{
+				const char *version;
+				const char *path_data;
+				espeak_Initialize(AUDIO_OUTPUT_SYNCHRONOUS, 0, data_path, espeakINITIALIZE_DONT_EXIT);
+				version = espeak_Info(&path_data);
+				printf("\n");
+				printf("eSpeak text-to-speech: %s  Data at: %s\n%s", version, path_data, help_text);
+				exit(0);
+			}
 			break;
 
 		case 'k':

src/extras.cpp

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2006 to 2010 by Jonathan Duddington                     *
+ *   Copyright (C) 2006 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -325,6 +325,7 @@ void Lexicon_It(int pass)
 	char buf_out[200];
 	char buf_error[200];
 	char last_listx[200];
+int test;
 	static const char *vowels1 = "aeiou";
 	static const char *vowels2 = "aeou";
 
@@ -455,6 +456,7 @@ void Lexicon_It(int pass)
 				break;
 			}
 		}
+
 		// translate
 		TranslateWord(translator,&word1[1],0, NULL);
 		DecodePhonemes(word_phonemes,phonemes);
@@ -953,7 +955,10 @@ void Lexicon_Bg()
 	}
 	input_length = GetFileLength(buf);
 
-	sprintf(fname,"%s%c%s",path_dsource,PATHSEP,"bg_listx_1");
+	sprintf(fname,"%s%c%s",path_dsource,PATHSEP,"bg_listx");
+	remove(fname);
+	CompileDictionary(path_dsource,"bg",NULL,NULL,0);
+
 	if((f_out = fopen(fname,"w")) == NULL)
 	{
 		wxLogError(_T("Can't write to: ")+wxString(fname,wxConvLocal));
@@ -961,8 +966,8 @@ void Lexicon_Bg()
 		return;
 	}
 
-	sprintf(fname,"%s%c%s",path_dsource,PATHSEP,"bg_log");
-	f_log = fopen(fname,"w");
+//	sprintf(fname,"%s%c%s",path_dsource,PATHSEP,"bg_log");
+//	f_log = fopen(fname,"w");
 
 	LoadVoice("bg",0);
 	progress = new wxProgressDialog(_T("Lexicon"),_T(""),input_length);
@@ -1089,11 +1094,13 @@ void Lexicon_Bg()
 
 	fclose(f_in);
 	fclose(f_out);
-	fclose(f_log);
+//	fclose(f_log);
+
+	CompileDictionary(path_dsource,"bg",NULL,NULL,0);
 
 
 	delete progress;
-	sprintf(buf,"Lexicon: Input %d,  Output %d,  Failed %d",n_words,n_out,n_wrong);
+	sprintf(buf,"Lexicon: Input %d,  Output %d,  $text %d",n_words,n_out,n_wrong);
 	wxLogStatus(wxString(buf,wxConvLocal));
 }  // end of Lexicon_Bg
 
@@ -1101,7 +1108,17 @@ void Lexicon_Bg()
 
 void Lexicon_Ru()
 {//==============
-	// compare stress markings in Russian RuLex file with lookup in ru_rules
+// compare stress markings in Russian RuLex file with lookup in ru_rules
+
+// Input file contains a list of Russian words (UTF-8), one per line
+// Stress position can be indicated either by:
+// A $ sign and syllable number after the word, eg:
+//   абажура  $3
+// or by a + sign after the stressed vowel, eg:
+//   абажу+ра
+
+// espeakedit produces a file:  dictsource/ru_listx  and a log file dictsource/ru_log
+
 	int ix;
 	char *p;
 	int  c;
@@ -1110,12 +1127,14 @@ void Lexicon_Ru()
 	FILE *f_log;
 	FILE *f_roots;
 	PHONEME_TAB *ph;
+	int ph_code;
 	int vcount;
 	int ru_stress;
 	int max_stress;
 	int max_stress_posn;
 	int n_words=0;
 	int n_wrong=0;
+	int n_errors=0;
 	int wlength;
 	int input_length;
 
@@ -1125,9 +1144,6 @@ void Lexicon_Ru()
 	int len;
 	int check_root;
 
-	int *p_unicode;
-	int unicode[80];
-
 	char word[80];
 	char word2[80];
 	int counts[20][20][10];
@@ -1136,8 +1152,8 @@ void Lexicon_Ru()
 	char buf[200];
 	char fname[sizeof(path_dsource)+20];
 
-	// KOI8-R codes for Russian vowels
-	static unsigned char vowels[] = {0xa3,0xc0,0xc1,0xc5,0xc9,0xcf,0xd1,0xd5,0xd9,0xdc,0};
+	// character codes for Russian vowels
+	static unsigned short ru_vowels[] = {0x430,0x435,0x438,0x439,0x43e,0x443,0x44d,0x44e,0x44f,0x450,0x451,0};
 
 	typedef struct {
 		const char *suffix;
@@ -1190,7 +1206,12 @@ void Lexicon_Ru()
 	}
 	input_length = GetFileLength(buf);
 
-	sprintf(fname,"%s%c%s",path_dsource,PATHSEP,"ru_listx_1");
+	sprintf(fname,"%s%c%s",path_dsource,PATHSEP,"ru_listx");
+	remove(fname);
+
+	// compile ru_dict without ru_listx
+	CompileDictionary(path_dsource,"ru",NULL,NULL,0);
+
 	if((f_out = fopen(fname,"w")) == NULL)
 	{
 		wxLogError(_T("Can't write to: ")+wxString(fname,wxConvLocal));
@@ -1201,7 +1222,7 @@ void Lexicon_Ru()
 	sprintf(fname,"%s%c%s",path_dsource,PATHSEP,"ru_log");
 	f_log = fopen(fname,"w");
 	sprintf(fname,"%s%c%s",path_dsource,PATHSEP,"ru_roots_1");
-	f_roots = fopen(fname,"w");
+//	f_roots = fopen(fname,"w");
 
 	LoadVoice("ru",0);
 
@@ -1221,46 +1242,46 @@ void Lexicon_Ru()
 		if(fgets(buf,sizeof(buf),f_in) == NULL)
 			break;
 
-		if(isspace2(buf[0]))
-			continue;
+		if((p = strstr(buf,"//")) != NULL)
+			*p = '\n';   // truncate at comment
 
-		// convert word from KOI8-R to UTF8
 		p = buf;
+		while((*p == ' ') || (*p == '\t')) p++;
+		if(*p == '\n')
+			continue;  // blank line
+
 		ix = 0;
 		wlength = 0;
-p_unicode = unicode;
-		while(!isspace2(c = (*p++ & 0xff)))
+		vcount = 0;
+		ru_stress = -1;
+
+		for(;;)
 		{
-			if(c >= 0xa0)
-			{
-				c = KOI8_R[c-0xa0];
-*p_unicode++ = c;
-			}
+			p += utf8_in(&c, p);
+			if(isspace(c))
+				break;
 
-
-			wlength++;
-			ix += utf8_out(c,&word[ix]);
+			if(lookupwchar(ru_vowels, c))
+				vcount++;
+			if(c == '+')
+				ru_stress = vcount;
+			else
+				ix += utf8_out(c, &word[ix]);
 		}
 		word[ix] = 0;
-*p_unicode=0;
 
-		sprintf(word2," %s ",word);
+		sprintf(word2," %s ",word);   // surround word by spaces before calling TranslateWord()
 
-		// find the marked stress position
-		vcount = 0;
-		ru_stress = 0;
-		while(*p == ' ') p++;
-
-		while((c = (*p++ & 0xff)) != '\n')
+		// find the marked stress position, if it has not been marked by a + after the stressed vowel
+		if(ru_stress == -1)
 		{
-			if(c == '+')
+			while((*p == ' ') || (*p == '\t')) p++;
+			sscanf(p,"$%d",&ru_stress);
+			if(ru_stress == -1)
 			{
-				ru_stress = vcount;
-				break;
-			}
-			if(strchr((char *)vowels,c) != NULL)
-			{
-				vcount++;
+				n_errors++;
+				fprintf(f_log,"%s",buf);
+				continue;   // stress position not found
 			}
 		}
 
@@ -1275,13 +1296,13 @@ p_unicode = unicode;
 		check_root = 0;
 
 		ph = phoneme_tab[phonPAUSE];
-		for(p=word_phonemes; *p != 0; p++)
+		for(p=word_phonemes; (ph_code = *p & 0xff) != 0; p++)
 		{
-			ph = phoneme_tab[(unsigned int)*p];
+			ph = phoneme_tab[ph_code];
 			if(ph == NULL)
 				continue;
 
-			if(ph->type == phVOWEL)
+			if((ph->type == phVOWEL) && (ph_code != phonSCHWA_SHORT))
 				vcount++;
 			if(ph->type == phSTRESS)
 			{
@@ -1300,6 +1321,7 @@ p_unicode = unicode;
 			{
 				fprintf(f_log,"%s\t $%d\t // %s\n",word,ru_stress,phonemes);
 			}
+			n_errors++;
 		}
 		else
 		{
@@ -1309,7 +1331,14 @@ p_unicode = unicode;
 			{
 				n_wrong++;
 				if((ru_stress==0) || (ru_stress > 7))
+				{
 					fprintf(f_out,"// ");   // we only have $1 to $7 to indicate stress position
+					if(f_log != NULL)
+					{
+						fprintf(f_log,"%s\t $%d\t // %s\n",word,ru_stress,phonemes);
+					}
+					n_errors++;
+				}
 				else
 					check_root = 1;
 
@@ -1329,6 +1358,8 @@ p_unicode = unicode;
 			}
 		}
 
+
+#ifdef deleted
 		if(check_root)
 		{
 			// does this word match any suffixes ?
@@ -1351,13 +1382,14 @@ p_unicode = unicode;
 				}
 			}
 		}
+#endif
 	}
 
 	fclose(f_in);
 	fclose(f_out);
-	fclose(f_roots);
+//	fclose(f_roots);
 
-	sprintf(buf,"Lexicon: Total %d  OK %d  wrong %d",n_words,n_words-n_wrong,n_wrong);
+	sprintf(buf,"Lexicon: Total %d  OK %d  fixed %d  errors %d (see ru_log)",n_words,n_words-n_wrong,n_wrong,n_errors);
 	if(gui_flag)
 	{
 		delete progress;
@@ -1374,13 +1406,13 @@ p_unicode = unicode;
 #ifdef deleted
 		// list tables of frequency of stress position for words of different syllable lengths
 		int j,k;
-		for(ix=0; ix<12; ix++)
+		for(ix=2; ix<12; ix++)
 		{
-			fprintf(f_log,"%2d syl: ",ix);
+			fprintf(f_log,"%2d syllables\n",ix);
 			for(k=0; k<10; k++)
 			{
 				fprintf(f_log,"  %2d :",k);
-				for(j=0; j<10; j++)
+				for(j=1; j<=ix; j++)
 				{
 					fprintf(f_log,"%6d ",counts[ix][j][k]);
 				}
@@ -2273,7 +2305,8 @@ if(control==2)
 	espeak_SetSynthCallback(TestSynthCallback);
 	espeak_SetUriCallback(TestUriCallback);
 
-  espeak_Synth(textbuf, strlen(textbuf)+1, 0, POS_CHARACTER, 0,  espeakSSML|espeakCHARS_UTF8, &unique_identifier, (void *)user_data);
+	espeak_SetVoiceByName("en");
+	espeak_Synth(textbuf, strlen(textbuf)+1, 0, POS_CHARACTER, 0,  espeakSSML|espeakCHARS_UTF8, &unique_identifier, (void *)user_data);
 //  espeak_Synth(text1, strlen(text1)+1, 0, POS_CHARACTER, 0,  espeakSSML|espeakCHARS_UTF8, &unique_identifier, (void *)(user_data+1));
 
   espeak_SetParameter(espeakPUNCTUATION, 1, 0);

src/menus.cpp

@@ -123,7 +123,7 @@ wxMenuBar *MakeMenu(int type, const char *dict_name)
 	vowelchart_menu->Append(MENU_VOWELCHART3, _("From directory of vowel files..."));
 
 	lexicon_menu = new wxMenu;
-	lexicon_menu->Append(MENU_LEXICON_RU, _("Russian (lexicon.dict)"));
+	lexicon_menu->Append(MENU_LEXICON_RU, _("Russian"));
 	lexicon_menu->Append(MENU_LEXICON_BG, _("Bulgarian"));
 	lexicon_menu->Append(MENU_LEXICON_DE, _("German"));
 	lexicon_menu->Append(MENU_LEXICON_IT, _("Italian"));

src/numbers.cpp

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2005 to 2010 by Jonathan Duddington                     *
+ *   Copyright (C) 2005 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -768,7 +768,7 @@ static int CheckDotOrdinal(Translator *tr, char *word, char *word_end, WORD_TAB
 			else
 				utf8_in(&c2, &word_end[0]);
 
-			if((word_end[1] != 0) && ((c2 == 0) || (wtab[0].flags & FLAG_COMMA_AFTER) || IsAlpha(c2)))
+			if((word_end[0] != 0) && (word_end[1] != 0) && ((c2 == 0) || (wtab[0].flags & FLAG_COMMA_AFTER) || IsAlpha(c2)))
 			{
 				// ordinal number is indicated by dot after the number
 				// but not if the next word starts with an upper-case letter
@@ -1089,7 +1089,7 @@ static int LookupThousands(Translator *tr, int value, int thousandplex, int thou
 		return(1);
 
 	return(found_value);
-}
+}  // end f LookupThousands
 
 
 static int LookupNum2(Translator *tr, int value, int control, char *ph_out)
@@ -1395,7 +1395,15 @@ static int LookupNum3(Translator *tr, int value, char *ph_out, int suppress_null
 		}
 		if(found == 0)
 		{
-			Lookup(tr, "_0C", ph_100);
+			if(tensunits==0)
+			{
+				// special form for exact hundreds?
+				found = Lookup(tr, "_0C0", ph_100);
+			}
+			if(!found)
+			{
+				Lookup(tr, "_0C", ph_100);
+			}
 		}
 
 		if(((tr->langopts.numbers & NUM_1900) != 0) && (hundreds == 19))
@@ -1671,13 +1679,10 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 				{
 					// this is an ordinal suffix
 					ordinal = 2;
+					flags[0] |= FLAG_SKIPWORDS;
+					skipwords = 1; 
 				}
 			}
-			if(ordinal)
-			{
-				flags[0] |= FLAG_SKIPWORDS;
-				skipwords = 1;
-			}
 		}
 	}
 

src/phoneme.h

@@ -54,8 +54,9 @@
 #define phNONSYLLABIC  0x100000   // don't count this vowel as a syllable when finding the stress position 
 #define phLONG         0x200000
 #define phLENGTHENSTOP 0x400000  // make the pre-pause slightly longer
-#define phRHOTIC       0x800000
+#define phRHOTIC       0x800000  // bit 23
 #define phNOPAUSE     0x1000000
+#define phPREVOICE    0x2000000  // for voiced stops
 
 #define phFLAG1      0x10000000
 #define phFLAG2      0x20000000

src/phonemelist.cpp

@@ -131,7 +131,7 @@ void MakePhonemeList(Translator *tr, int post_pause, int start_sentence)
 	PHONEME_LIST *plist3_inserted = NULL;
 	PHONEME_LIST ph_list3[N_PHONEME_LIST];
 
-	static PHONEME_LIST2 ph_list2_null = {0,0,0,0,0};
+	static PHONEME_LIST2 ph_list2_null = {0,0,0,0,0,0};
 	PHONEME_LIST2 *plist2 = &ph_list2_null;
 
 	plist2 = ph_list2;

src/readclause.cpp

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2005 to 2010 by Jonathan Duddington                     *
+ *   Copyright (C) 2005 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -101,6 +101,7 @@ static const unsigned short punct_chars[] = {',','.','?','!',':',';',
   0x1366,
   0x1367,
   0x1368,
+  0x10fb,  // Georgian paragraph
 
   0x3001,  // ideograph comma
   0x3002,  // ideograph period
@@ -142,7 +143,8 @@ static const unsigned int punct_attributes [] = { 0,
   CLAUSE_COLON,      // Ethiopic colon
   CLAUSE_COLON,      // Ethiopic preface colon
   CLAUSE_QUESTION,   // Ethiopic question mark
-  CLAUSE_PERIOD,     // Ethiopic paragraph
+  CLAUSE_PARAGRAPH,     // Ethiopic paragraph
+  CLAUSE_PARAGRAPH,     // Georgian paragraph
 
   CLAUSE_COMMA+0x8000,      // ideograph comma
   CLAUSE_PERIOD+0x8000,     // ideograph period
@@ -884,7 +886,7 @@ static int AnnouncePunctuation(Translator *tr, int c1, int *c2_ptr, char *output
 			if(punct_count==1)
 			{
 //				sprintf(buf,"%s %s %s",tone_punct_on,punctname,tone_punct_off);
-				sprintf(buf,"%s",punctname);
+				sprintf(buf," %s",punctname);   // we need the space before punctname, to ensure it doesn't merge with the previous word  (eg.  "2.-a")
 			}
 			else
 			if(punct_count < 4)
@@ -900,7 +902,7 @@ static int AnnouncePunctuation(Translator *tr, int c1, int *c2_ptr, char *output
 			}
 			else
 			{
-				sprintf(buf,"%s %d %s",
+				sprintf(buf," %s %d %s",
 						punctname,punct_count,punctname);
 			}
 		}
@@ -1193,7 +1195,7 @@ static wchar_t *GetSsmlAttribute(wchar_t *pw, const char *name)
 				while(iswspace(*pw)) pw++;
 				if(*pw == '=') pw++;
 				while(iswspace(*pw)) pw++;
-				if(*pw == '"')
+				if((*pw == '"') || (*pw == '\''))  // allow single-quotes ?
 					return(pw+1);
 				else
 					return(empty);
@@ -1215,7 +1217,7 @@ static int attrcmp(const wchar_t *string1, const char *string2)
 	for(ix=0; (string1[ix] == string2[ix]) && (string1[ix] != 0); ix++)
 	{
 	}
-	if((string1[ix]=='"') && (string2[ix]==0))
+	if(((string1[ix]=='"') || (string1[ix]=='\'')) && (string2[ix]==0))
 		return(0);
 	return(1);
 }
@@ -1284,7 +1286,7 @@ static int attr_prosody_value(int param_type, const wchar_t *pw, int *value_out)
 {//=============================================================================
 	int sign = 0;
 	wchar_t *tail;
-	float value;
+	double value;
 
 	while(iswspace(*pw)) pw++;
 	if(*pw == '+')
@@ -1297,7 +1299,7 @@ static int attr_prosody_value(int param_type, const wchar_t *pw, int *value_out)
 		pw++;	
 		sign = -1;
 	}
-	value = (float)wcstod(pw,&tail);
+	value = (double)wcstod(pw,&tail);
 	if(tail == pw)
 	{
 		// failed to find a number, return 100%
@@ -1324,7 +1326,10 @@ static int attr_prosody_value(int param_type, const wchar_t *pw, int *value_out)
 
 	if(param_type == espeakRATE)
 	{
-		*value_out = (int)(value * 100);
+		if(sign == 0)
+			*value_out = (int)(value * 100);
+		else
+			*value_out = 100 + (int)(sign * value * 100);
 		return(2);   // percentage
 	}
 

src/setlengths.cpp

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2005 to 2007 by Jonathan Duddington                     *
+ *   Copyright (C) 2005 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -31,6 +31,7 @@
 #include "translate.h"
 
 extern int GetAmplitude(void);
+extern void DoSonicSpeed(int value);
 
 
 // convert from words-per-minute to internal speed factor
@@ -140,27 +141,54 @@ void SetSpeed(int control)
 	int s1;
 	int wpm;
 	int wpm2;
+	double sonic;
 
 	speed.loud_consonants = 0;
 	speed.min_sample_len = 450;
 	speed.lenmod_factor = 110;   // controls the effect of FRFLAG_LEN_MOD reduce length change
 	speed.lenmod2_factor = 100;
+	speed.min_pause = 5;
 
 	wpm = embedded_value[EMBED_S];
 	if(control == 2)
 		wpm = embedded_value[EMBED_S2];
 
-	// fast_settings[0]  wpm where speedup starts
-	// fast_settings[1]  wpm for maximum espeak speed
-	// fast_settings[2]  maximum espeak speed
+	if(voice->speed_percent > 0)
+	{
+		wpm = (wpm * voice->speed_percent)/100;
+	}
 
-	sonicSpeed = 1.0;
-	if(wpm > speed.fast_settings[0])
+	if(control & 2)
+	{
+		DoSonicSpeed(1 * 1024);
+	}
+	if((wpm > 450) || ((wpm > speed.fast_settings[0]) && (wpm > 350)))
 	{
 		wpm2 = wpm;
-		wpm = speed.fast_settings[2];
-		// Maybe we can do better than a 10 wpm fudge factor?
-		sonicSpeed = ((double)wpm2 + 10)/wpm;
+		wpm = 175;
+
+		// set special eSpeak speed parameters for Sonic use
+		// The eSpeak output will be speeded up by at least x2
+		x = 73;
+		if(control & 1)
+		{
+			speed1 = (x * voice->speedf1)/256;
+			speed2 = (x * voice->speedf2)/256;
+			speed3 = (x * voice->speedf3)/256;
+		}
+		if(control & 2)
+		{
+			sonic = ((double)wpm2)/wpm;
+			DoSonicSpeed((int)(sonic * 1024));
+			speed.pause_factor = 85;
+			speed.clause_pause_factor = 80;
+			speed.min_pause = 22;
+			speed.min_sample_len = 450*2;
+			speed.wav_factor = 211;
+			speed.lenmod_factor = 210;
+			speed.lenmod2_factor = 170;
+		}
+		return;
 	}
 
 
@@ -173,10 +201,6 @@ void SetSpeed(int control)
 	}
 #endif
 
-	if(voice->speed_percent > 0)
-	{
-		wpm = (wpm * voice->speed_percent)/100;
-	}
 	if(wpm > 450)
 		wpm = 450;
 

src/speak.cpp

@@ -542,6 +542,9 @@ int main (int argc, char **argv)
 
 		if(c == '-')
 		{
+			if(p[0] == 0)
+				break;   // -- means don't interpret further - as commands
+
 			opt_string="";
 			for(ix=0; ;ix++)
 			{
@@ -593,7 +596,8 @@ int main (int argc, char **argv)
 			break;
 
 		case 'h':
-			printf("\nspeak text-to-speech: %s\n%s",version_string,help_text);
+			init_path(argv[0],data_path);
+			printf("\nspeak text-to-speech: %s   Data at: %s\n%s",version_string,path_home,help_text);
 			exit(0);
 			break;
 

src/speak_lib.cpp

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2005 to 2010 by Jonathan Duddington                     *
+ *   Copyright (C) 2005 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -370,8 +370,8 @@ static void init_path(const char *path)
 #endif
 }
 
-static int initialise(void)
-{//========================
+static int initialise(int control)
+{//===============================
 	int param;
 	int result;
 
@@ -382,7 +382,10 @@ static int initialise(void)
 		if(result == -1)
 		{
 			fprintf(stderr,"Failed to load espeak-data\n");
-			exit(1);
+			if((control & espeakINITIALIZE_DONT_EXIT) == 0)
+			{
+				exit(1);
+			}
 		}
 		else
 			fprintf(stderr,"Wrong version of espeak-data 0x%x (expects 0x%x) at %s\n",result,version_phdata,path_home);
@@ -745,7 +748,7 @@ ENTER("espeak_Initialize");
 #endif
 	
 	init_path(path);
-	initialise();
+	initialise(options);
 	select_output(output_type);
 
 	if(f_logespeak)
@@ -1222,8 +1225,12 @@ ESPEAK_API espeak_ERROR espeak_Terminate(void)
 	return EE_OK;
 }   //  end of espeak_Terminate
 
-ESPEAK_API const char *espeak_Info(void *)
-{//=======================================
+ESPEAK_API const char *espeak_Info(const char **ptr)
+{//=================================================
+	if(ptr != NULL)
+	{
+		*ptr = path_home;
+	}
 	return(version_string);
 }
 

src/speak_lib.h

@@ -28,6 +28,12 @@
 #include <stdio.h>
 #include <stddef.h>
 
+#ifdef __WIN32__
+#define ESPEAK_API __declspec(dllexport)
+#else
+#define ESPEAK_API
+#endif
+
 #define ESPEAK_API_REVISION  6
 /*
 Revision 2
@@ -157,11 +163,13 @@ typedef enum {
 	EE_NOT_FOUND=2
 } espeak_ERROR;
 
+#define espeakINITIALIZE_PHONEME_EVENTS 0x0001
+#define espeakINITIALIZE_DONT_EXIT     0x8000
 
 #ifdef __cplusplus
 extern "C"
 #endif
-int espeak_Initialize(espeak_AUDIO_OUTPUT output, int buflength, const char *path, int options);
+ESPEAK_API int espeak_Initialize(espeak_AUDIO_OUTPUT output, int buflength, const char *path, int options);
 /* Must be called before any synthesis functions are called.
    output: the audio data can either be played by eSpeak or passed back by the SynthCallback function.
 
@@ -169,8 +177,8 @@ int espeak_Initialize(espeak_AUDIO_OUTPUT output, int buflength, const char *pat
 
    path: The directory which contains the espeak-data directory, or NULL for the default location.
 
-   options: bit 0: 1=allow espeakEVENT_PHONEME events.
-
+   options: bit 0:  1=allow espeakEVENT_PHONEME events.
+            bit 15: 1=don't exit if espeak_data is not found (used for --help)
 
    Returns: sample rate in Hz, or -1 (EE_INTERNAL_ERROR).
 */
@@ -180,7 +188,7 @@ typedef int (t_espeak_callback)(short*, int, espeak_EVENT*);
 #ifdef __cplusplus
 extern "C"
 #endif
-void espeak_SetSynthCallback(t_espeak_callback* SynthCallback);
+ESPEAK_API void espeak_SetSynthCallback(t_espeak_callback* SynthCallback);
 /* Must be called before any synthesis functions are called.
    This specifies a function in the calling program which is called when a buffer of
    speech sound data has been produced. 
@@ -208,7 +216,7 @@ int SynthCallback(short *wav, int numsamples, espeak_EVENT *events);
 #ifdef __cplusplus
 extern "C"
 #endif
-void espeak_SetUriCallback(int (*UriCallback)(int, const char*, const char*));
+ESPEAK_API void espeak_SetUriCallback(int (*UriCallback)(int, const char*, const char*));
 /* This function may be called before synthesis functions are used, in order to deal with
    <audio> tags.  It specifies a callback function which is called when an <audio> element is
    encountered and allows the calling program to indicate whether the sound file which
@@ -249,7 +257,7 @@ int UriCallback(int type, const char *uri, const char *base);
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_Synth(const void *text,
+ESPEAK_API espeak_ERROR espeak_Synth(const void *text,
 	size_t size,
 	unsigned int position,
 	espeak_POSITION_TYPE position_type,
@@ -305,7 +313,7 @@ espeak_ERROR espeak_Synth(const void *text,
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_Synth_Mark(const void *text,
+ESPEAK_API espeak_ERROR espeak_Synth_Mark(const void *text,
 	size_t size,
 	const char *index_mark,
 	unsigned int end_position,
@@ -329,7 +337,7 @@ espeak_ERROR espeak_Synth_Mark(const void *text,
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_Key(const char *key_name);
+ESPEAK_API espeak_ERROR espeak_Key(const char *key_name);
 /* Speak the name of a keyboard key.
    If key_name is a single character, it speaks the name of the character.
    Otherwise, it speaks key_name as a text string.
@@ -343,7 +351,7 @@ espeak_ERROR espeak_Key(const char *key_name);
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_Char(wchar_t character);
+ESPEAK_API espeak_ERROR espeak_Char(wchar_t character);
 /* Speak the name of the given character 
 
    Return: EE_OK: operation achieved 
@@ -388,7 +396,7 @@ typedef enum {
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_SetParameter(espeak_PARAMETER parameter, int value, int relative);
+ESPEAK_API espeak_ERROR espeak_SetParameter(espeak_PARAMETER parameter, int value, int relative);
 /* Sets the value of the specified parameter.
    relative=0   Sets the absolute value of the parameter.
    relative=1   Sets a relative value of the parameter.
@@ -425,7 +433,7 @@ espeak_ERROR espeak_SetParameter(espeak_PARAMETER parameter, int value, int rela
 #ifdef __cplusplus
 extern "C"
 #endif
-int espeak_GetParameter(espeak_PARAMETER parameter, int current);
+ESPEAK_API int espeak_GetParameter(espeak_PARAMETER parameter, int current);
 /* current=0  Returns the default value of the specified parameter.
    current=1  Returns the current value of the specified parameter, as set by SetParameter()
 */
@@ -433,7 +441,7 @@ int espeak_GetParameter(espeak_PARAMETER parameter, int current);
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_SetPunctuationList(const wchar_t *punctlist);
+ESPEAK_API espeak_ERROR espeak_SetPunctuationList(const wchar_t *punctlist);
 /* Specified a list of punctuation characters whose names are to be spoken when the
    value of the Punctuation parameter is set to "some".
 
@@ -448,7 +456,7 @@ espeak_ERROR espeak_SetPunctuationList(const wchar_t *punctlist);
 #ifdef __cplusplus
 extern "C"
 #endif
-void espeak_SetPhonemeTrace(int value, FILE *stream);
+ESPEAK_API void espeak_SetPhonemeTrace(int value, FILE *stream);
 /* Controls the output of phoneme symbols for the text
    value=0  No phoneme output (default)
    value=1  Output the translated phoneme symbols for the text
@@ -461,7 +469,7 @@ void espeak_SetPhonemeTrace(int value, FILE *stream);
 #ifdef __cplusplus
 extern "C"
 #endif
-void espeak_CompileDictionary(const char *path, FILE *log, int flags);
+ESPEAK_API void espeak_CompileDictionary(const char *path, FILE *log, int flags);
 /* Compile pronunciation dictionary for a language which corresponds to the currently
    selected voice.  The required voice should be selected before calling this function.
 
@@ -514,7 +522,7 @@ typedef struct {
 #ifdef __cplusplus
 extern "C"
 #endif
-const espeak_VOICE **espeak_ListVoices(espeak_VOICE *voice_spec);
+ESPEAK_API const espeak_VOICE **espeak_ListVoices(espeak_VOICE *voice_spec);
 /* Reads the voice files from espeak-data/voices and creates an array of espeak_VOICE pointers.
    The list is terminated by a NULL pointer
 
@@ -526,7 +534,7 @@ const espeak_VOICE **espeak_ListVoices(espeak_VOICE *voice_spec);
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_SetVoiceByName(const char *name);
+ESPEAK_API espeak_ERROR espeak_SetVoiceByName(const char *name);
 /* Searches for a voice with a matching "name" field.  Language is not considered.
    "name" is a UTF8 string.
 
@@ -539,7 +547,7 @@ espeak_ERROR espeak_SetVoiceByName(const char *name);
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_SetVoiceByProperties(espeak_VOICE *voice_spec);
+ESPEAK_API espeak_ERROR espeak_SetVoiceByProperties(espeak_VOICE *voice_spec);
 /* An espeak_VOICE structure is used to pass criteria to select a voice.  Any of the following
    fields may be set:
 
@@ -559,7 +567,7 @@ espeak_ERROR espeak_SetVoiceByProperties(espeak_VOICE *voice_spec);
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_VOICE *espeak_GetCurrentVoice(void);
+ESPEAK_API espeak_VOICE *espeak_GetCurrentVoice(void);
 /* Returns the espeak_VOICE data for the currently selected voice.
    This is not affected by temporary voice changes caused by SSML elements such as <voice> and <s>
 */
@@ -567,7 +575,7 @@ espeak_VOICE *espeak_GetCurrentVoice(void);
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_Cancel(void);
+ESPEAK_API espeak_ERROR espeak_Cancel(void);
 /* Stop immediately synthesis and audio output of the current text. When this
    function returns, the audio output is fully stopped and the synthesizer is ready to
    synthesize a new message.
@@ -580,14 +588,14 @@ espeak_ERROR espeak_Cancel(void);
 #ifdef __cplusplus
 extern "C"
 #endif
-int espeak_IsPlaying(void);
+ESPEAK_API int espeak_IsPlaying(void);
 /* Returns 1 if audio is played, 0 otherwise.
 */
 
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_Synchronize(void);
+ESPEAK_API espeak_ERROR espeak_Synchronize(void);
 /* This function returns when all data have been spoken.
    Return: EE_OK: operation achieved 
 	   EE_INTERNAL_ERROR.
@@ -596,7 +604,7 @@ espeak_ERROR espeak_Synchronize(void);
 #ifdef __cplusplus
 extern "C"
 #endif
-espeak_ERROR espeak_Terminate(void);
+ESPEAK_API espeak_ERROR espeak_Terminate(void);
 /* last function to be called.
    Return: EE_OK: operation achieved 
 	   EE_INTERNAL_ERROR.
@@ -606,8 +614,8 @@ espeak_ERROR espeak_Terminate(void);
 #ifdef __cplusplus
 extern "C"
 #endif
-const char *espeak_Info(void* ptr);
+ESPEAK_API const char *espeak_Info(const char **path_data);
 /* Returns the version number string.
-   The parameter is for future use, and should be set to NULL
+   path_data  returns the path to espeak_data
 */
 #endif

src/speech.h

@@ -17,6 +17,8 @@
  *               <http://www.gnu.org/licenses/>.                           *
  ***************************************************************************/
 
+#ifndef SPEECH_H
+#define SPEECH_H
 
 #include <sys/types.h>
 
@@ -41,7 +43,7 @@
 //#define USE_PULSEAUDIO
 #define USE_NANOSLEEP
 #define __cdecl 
-#define ESPEAK_API  extern "C"
+//#define ESPEAK_API  extern "C"
 
 #ifdef LIBRARY
 #define USE_ASYNC
@@ -85,3 +87,4 @@ int  GetFileLength(const char *filename);
 char *Alloc(int size);
 void Free(void *ptr);
 
+#endif // SPEECH_H

src/synth_mbrola.cpp

@@ -133,14 +133,18 @@ espeak_ERROR LoadMbrolaTable(const char *mbrola_voice, const char *phtrans, int
 
 	sprintf(path,"%s/mbrola/%s",path_home,mbrola_voice);
 #ifdef PLATFORM_POSIX
+	// if not found, then also look in
+	//   usr/share/mbrola/xx, /usr/share/mbrola/xx/xx, /usr/share/mbrola/voices/xx
 	if(GetFileLength(path) <= 0)
 	{
-		// mbrola voice file not found, look in /usr/share
      sprintf(path,"/usr/share/mbrola/%s",mbrola_voice);
 	}
 	if(GetFileLength(path) <= 0)
 	{
-		// mbrola voice file not found, look in /usr/share
+     sprintf(path,"/usr/share/mbrola/%s/%s",mbrola_voice,mbrola_voice);
+	}
+	if(GetFileLength(path) <= 0)
+	{
      sprintf(path,"/usr/share/mbrola/voices/%s",mbrola_voice);
 	}
 #endif

src/synthdata.cpp

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2005 to 2010 by Jonathan Duddington                     *
+ *   Copyright (C) 2005 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.44.19  20.Dec.10";
+const char *version_string = "1.44.42  12.Mar.11";
 const int version_phdata  = 0x014415;
 
 int option_device_number = -1;
@@ -496,7 +496,7 @@ void LoadConfig(void)
 		else
 		if(memcmp(buf,"pa_device",9)==0)
 		{
-			sscanf(&buf[7],"%d",&option_device_number);
+			sscanf(&buf[10],"%d",&option_device_number);
 		}
 		else
 		if(memcmp(buf,"soundicon",9)==0)
@@ -617,6 +617,13 @@ static bool InterpretCondition(Translator *tr, int control, PHONEME_LIST *plist,
 			if(plist[0].sourceix)
 				return(false);
 		}
+		if(which==7)
+		{
+			// nextPh2 not word boundary
+			if((plist[1].sourceix) || (plist[2].sourceix))
+				return(false);
+		}
+
 		if(which==6)
 		{
 			// nextVowel, not word boundary
@@ -628,16 +635,11 @@ static bool InterpretCondition(Translator *tr, int control, PHONEME_LIST *plist,
 					break;
 			}
 		}
-		if(which==7)
-		{
-			// nextPh2 not word boundary
-			if((plist[1].sourceix) || (plist[2].sourceix))
-				return(false);
-		}
 		else
 		{
 			which = ph_position[which];
 		}
+
 		plist_this = plist;
 		plist = &plist[which-1];
 

src/synthesize.cpp

@@ -200,7 +200,10 @@ int PauseLength(int pause, int control)
 	else
 		len = (pause * speed.wav_factor)/256;
 
-	if(len < 5) len = 5;      // mS, limit the amount to which pauses can be shortened
+	if(len < speed.min_pause)
+	{
+		len = speed.min_pause;      // mS, limit the amount to which pauses can be shortened
+	}
 	return(len);
 }
 
@@ -1215,6 +1218,15 @@ void DoMarker(int type, int char_posn, int length, int value)
 }  // end of DoMarker
 
 
+void DoSonicSpeed(int value)
+{//=========================
+// value, multiplier * 1024
+	wcmdq[wcmdq_tail][0] = WCMD_SONIC_SPEED;
+	wcmdq[wcmdq_tail][1] = value;
+	WcmdqInc();
+}  // end of DoSonicSpeed
+
+
 void DoVoiceChange(voice_t *v)
 {//===========================
 // allocate memory for a copy of the voice data, and free it in wavegenfill()
@@ -1459,7 +1471,7 @@ int Generate(PHONEME_LIST *phoneme_list, int *n_ph, int resume)
 				}
 			}
 
-			if((prev->type==phVOWEL) || (prev->ph->phflags & phVOWEL2))
+			if((prev->type==phVOWEL) || (prev->ph->phflags & phVOWEL2) || (ph->phflags & phPREVOICE))
 			{
 				// a period of voicing before the release
 				InterpretPhoneme(NULL, 0x01, p, &phdata);

src/synthesize.h

@@ -334,7 +334,7 @@ typedef struct {
 // phflags
 #define i_isSibilant   0x45    // bit 5 in phflags
 #define i_isPalatal    0x49    // bit 9 in phflags
-#define i_isRhotic     0x56    // bit 22 in phflags
+#define i_isRhotic     0x57    // bit 23 in phflags
 #define i_isFlag1      0x5c
 #define i_isFlag2      0x5d
 #define i_isFlag3      0x5e
@@ -362,6 +362,7 @@ typedef struct {
 typedef struct {
 	int pause_factor;
 	int clause_pause_factor;
+	int min_pause;
 	int wav_factor;
 	int lenmod_factor;
 	int lenmod2_factor;
@@ -448,6 +449,7 @@ extern unsigned char pitch_adjust_tab[MAX_PITCH_VALUE+1];
 #define WCMD_EMBEDDED 12
 #define WCMD_MBROLA_DATA 13
 #define WCMD_FMT_AMPLITUDE 14
+#define WCMD_SONIC_SPEED 15
 
 
 

src/tr_languages.cpp

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2005 to 2010 by Jonathan Duddington                     *
+ *   Copyright (C) 2005 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -40,21 +40,29 @@
 #define L_grc  0x677263   // grc  Ancient Greek
 #define L_jbo  0x6a626f   // jbo  Lojban
 #define L_pap  0x706170   // pap  Papiamento]
+#define L_qvi  0x717669   // qvi  Kichwa
 #define L_shs  0x736873   // shs  Shuswap / Secwepemctsin
 #define L_zhy  0x7a6879   // zhy
 
 // start of unicode pages for character sets
-#define OFFSET_GREEK  0x380
+#define OFFSET_GREEK    0x380
 #define OFFSET_CYRILLIC 0x420
 #define OFFSET_ARMENIAN 0x530
-#define OFFSET_ARABIC 0x600
+#define OFFSET_ARABIC   0x600
+#define OFFSET_THAANA   0x780      // Divehi/Maldives
 #define OFFSET_DEVANAGARI  0x900
-#define OFFSET_BENGALI 0x980
+#define OFFSET_BENGALI  0x980
 #define OFFSET_GURMUKHI 0xa00
-#define OFFSET_TAMIL  0xb80
-#define OFFSET_KANNADA 0xc80
+#define OFFSET_GUJARATI 0xa80
+#define OFFSET_ORIYA    0xb00
+#define OFFSET_TAMIL    0xb80
+#define OFFSET_TELUGU   0xc00
+#define OFFSET_KANNADA  0xc80
 #define OFFSET_MALAYALAM 0xd00
-#define OFFSET_KOREAN 0x1100
+#define OFFSET_SINHALA  0x0d80
+#define OFFSET_GEORGIAN 0x1080
+
+#define OFFSET_KOREAN   0x1100
 
 static void Translator_Russian(Translator *tr);
 
@@ -106,7 +114,7 @@ static Translator* NewTranslator(void)
 {//===================================
 	Translator *tr;
 	int ix;
-	static const unsigned char stress_amps2[] = {17,17, 20,20, 20,22, 22,20 };
+	static const unsigned char stress_amps2[] = {18,18, 20,20, 20,22, 22,20 };
 	static const short stress_lengths2[8] = {182,140, 220,220, 220,240, 260,280};
 	static const wchar_t empty_wstring[1] = {0};
 	static const wchar_t punct_in_word[2] = {'\'', 0};  // allow hyphen within words
@@ -334,6 +342,9 @@ Translator *SelectTranslator(const char *name)
 	static const unsigned char stress_amps_sk[8] = {17,17, 20,20, 20,22, 22,21 };
 	static const short stress_lengths_sk[8] = {190,190, 210,210, 0,0, 210,210};
 
+	static const short stress_lengths_ta[8] = {200, 200,  210, 210,  0, 0,  230, 230};
+	static const unsigned char stress_amps_ta[8] = {18,18, 18,18, 20,20, 22,22 };
+
 	// convert name string into a word of up to 4 characters, for the switch()
 	while(*name != 0)
 		name2 = (name2 << 8) + *name++;
@@ -446,14 +457,15 @@ Translator *SelectTranslator(const char *name)
 
 	case L('d','e'):
 		{
-			static const short stress_lengths_de[8] = {150,130, 200,200,  0, 0, 250,260};
+			static const short stress_lengths_de[8] = {150,130, 200,200,  0, 0, 270,270};
+			static const unsigned char stress_amps_de[] = {20,20, 20,20, 20,22, 22,20 };
+			SetupTranslator(tr, stress_lengths_de, stress_amps_de);
 			tr->langopts.stress_rule = STRESSPOSN_1L;
 			tr->langopts.word_gap = 0x8;   // don't use linking phonemes
 			tr->langopts.vowel_pause = 0x30;
 			tr->langopts.param[LOPT_PREFIXES] = 1;
 			tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 0x10;  // devoice at end of word
 			tr->langopts.param[LOPT_LONG_VOWEL_THRESHOLD] = 175/2;
-			memcpy(tr->stress_lengths,stress_lengths_de,sizeof(tr->stress_lengths));
 		
 			tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SWAP_TENS | NUM_OMIT_1_HUNDRED | NUM_OMIT_1_THOUSAND | NUM_ALLOW_SPACE | NUM_ORDINAL_DOT | NUM_ROMAN;
 			SetLetterVowel(tr,'y');
@@ -461,6 +473,20 @@ Translator *SelectTranslator(const char *name)
 		}
 		break;
 
+	case L('d','v'):   // Divehi (Maldives)
+		{
+			SetupTranslator(tr,stress_lengths_ta,stress_amps_ta);
+			tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1;   // disable check for unpronouncable words
+			tr->langopts.length_mods0 = tr->langopts.length_mods;  // don't lengthen vowels in the last syllable
+			tr->letter_bits_offset = OFFSET_THAANA;
+			tr->langopts.stress_rule = STRESSPOSN_1L;
+			tr->langopts.stress_flags =  0x10004;   // use 'diminished' for unstressed final syllable
+			SetLetterBitsRange(tr,LETTERGP_B,0x26,0x30);   // vowel signs, and virama
+			tr->langopts.break_numbers = 0x14a8;  // 1000, 100,000  10,000,000 
+			tr->langopts.numbers = 1;
+		}
+		break;
+
 	case L('e','n'):
 		{
 			static const short stress_lengths_en[8] = {182,140, 220,220, 0,0, 248,275};
@@ -522,7 +548,7 @@ Translator *SelectTranslator(const char *name)
 
 	case L('e','o'):
 		{
-			static const short stress_lengths_eo[8] = {145, 145,  230, 170,    0,   0,  360, 370};
+			static const short stress_lengths_eo[8] = {150, 150,  230, 180,    0,   0,  300, 320};
 			static const unsigned char stress_amps_eo[] = {16,14, 20,20, 20,22, 22,21 };
 			static const wchar_t eo_char_apostrophe[2] = {'l',0};
 		
@@ -531,11 +557,11 @@ Translator *SelectTranslator(const char *name)
 			tr->charset_a0 = charsets[3];  // ISO-8859-3
 			tr->char_plus_apostrophe = eo_char_apostrophe;
 
-			tr->langopts.word_gap = 1;
+//			tr->langopts.word_gap = 1;
 			tr->langopts.vowel_pause = 2;
 			tr->langopts.stress_rule = STRESSPOSN_2R;
 			tr->langopts.stress_flags =  0x6 | 0x10; 
-			tr->langopts.unstressed_wd1 = 3;
+//			tr->langopts.unstressed_wd1 = 3;
 			tr->langopts.unstressed_wd2 = 2;
 
 			tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_OMIT_1_HUNDRED | NUM_ALLOW_SPACE | NUM_ROMAN;
@@ -583,7 +609,6 @@ Translator *SelectTranslator(const char *name)
 		}
 		break;
 
-
 	case L('e','u'):  // basque
 		{
 			static const short stress_lengths_eu[8] = {200, 200,  200, 200,  0, 0,  210, 230};  // very weak stress
@@ -604,6 +629,9 @@ Translator *SelectTranslator(const char *name)
 		break;
 
 
+	case L('e','t'):   // Estonian
+			tr->charset_a0 = charsets[4];   // ISO-8859-4
+		// drop through to Finnish
 	case L('f','i'):   // Finnish
 		{
 			static const unsigned char stress_amps_fi[8] = {18,16, 22,22, 20,22, 22,22 };
@@ -824,6 +852,26 @@ SetLengthMods(tr,3);  // all equal
 		}
 		break;
 
+	case L('k','a'):   // Georgian
+		{
+			// character codes offset by 0x1080
+			static const char ka_vowels[] = {0x50,0x54,0x58,0x5d,0x63,0x75,0x77,0};
+			static const char ka_consonants[] = {0x51,0x52,0x53,0x55,0x56,0x57,0x59,0x5a,0x5b,0x5c,0x5e,0x5f,0x60,0x61,0x62,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,0x6f,0x70,0x71,0x72,0x73,0x74,0x76,0};
+			SetupTranslator(tr,stress_lengths_ta,stress_amps_ta);
+			memset(tr->letter_bits,0,sizeof(tr->letter_bits));
+			SetLetterBits(tr,LETTERGP_A,ka_vowels);
+			SetLetterBits(tr,LETTERGP_C,ka_consonants);
+			SetLetterBits(tr,LETTERGP_VOWEL2,ka_vowels);
+
+			tr->langopts.stress_rule = STRESSPOSN_1L;
+			tr->langopts.stress_flags = S_FINAL_NO_2;
+			tr->letter_bits_offset = OFFSET_GEORGIAN;
+//			tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1;   // disable check for unpronouncable words
+			tr->langopts.max_initial_consonants = 7;
+			tr->langopts.numbers = NUM_VIGESIMAL | NUM_AND_UNITS | NUM_OMIT_1_HUNDRED |NUM_OMIT_1_THOUSAND | NUM_DFRACTION_5;
+		}
+		break;
+
 	case L('k','o'):   // Korean, TEST
 		{
 			static const char ko_ivowels[] = {0x63,0x64,0x67,0x68,0x6d,0x72,0x74,0x75,0};  // y and i vowels
@@ -913,7 +961,7 @@ SetLengthMods(tr,3);  // all equal
 			tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 0x10;  // devoice at end of word
 			SetLetterVowel(tr,'y');
 
-			tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SWAP_TENS | NUM_OMIT_1_HUNDRED | NUM_ALLOW_SPACE | NUM_1900 | NUM_ORDINAL_DOT;
+			tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SWAP_TENS | NUM_OMIT_1_HUNDRED | NUM_OMIT_1_THOUSAND | NUM_ALLOW_SPACE | NUM_1900 | NUM_ORDINAL_DOT;
 			tr->langopts.ordinal_indicator = "e";
 			tr->langopts.stress_flags = S_FIRST_PRIMARY;
 			memcpy(tr->stress_lengths,stress_lengths_nl,sizeof(tr->stress_lengths));
@@ -1083,6 +1131,7 @@ SetLengthMods(tr,3);  // all equal
 		break;
 
 	case L('s','w'):  // Swahili
+	case L('t','n'):  // Setswana
 		{
 			static const short stress_lengths_sw[8] = {160, 170,  200, 200,    0,   0,  320, 340};
 			static const unsigned char stress_amps_sw[] = {16,12, 19,19, 20,22, 22,21 };
@@ -1103,15 +1152,14 @@ SetLengthMods(tr,3);  // all equal
 	case L('m','l'):  // Malayalam
 	case L('k','n'):  // Kannada
 	case L('m','r'):  // Marathi
+	case L('t','e'):  // Telugu
 		{
-			static const short stress_lengths_ta[8] = {200, 200,  210, 210,  0, 0,  230, 230};
-			static const unsigned char stress_amps_ta[8] = {18,18, 18,18, 20,20, 22,22 };
-
 			SetupTranslator(tr,stress_lengths_ta,stress_amps_ta);
 			tr->langopts.length_mods0 = tr->langopts.length_mods;  // don't lengthen vowels in the last syllable
 
 			tr->langopts.stress_rule = STRESSPOSN_1L;
 			tr->langopts.stress_flags =  0x10004;   // use 'diminished' for unstressed final syllable
+			tr->langopts.spelling_stress = 1;
 			tr->langopts.break_numbers = 0x14a8;  // 1000, 100,000  10,000,000 
 
 			if(name2 == L('t','a'))
@@ -1127,6 +1175,7 @@ SetLengthMods(tr,3);  // all equal
 			if(name2 == L('m','l'))
 			{
 				tr->letter_bits_offset = OFFSET_MALAYALAM;
+				tr->langopts.numbers = NUM_OMIT_1_THOUSAND;
 			}
 			else
 			if(name2 == L('k','n'))
@@ -1134,6 +1183,12 @@ SetLengthMods(tr,3);  // all equal
 				tr->letter_bits_offset = OFFSET_KANNADA;
 				tr->langopts.numbers = 0x1;
 			}
+			else
+			if(name2 == L('t','e'))
+			{
+				tr->letter_bits_offset = OFFSET_TELUGU;
+				tr->langopts.numbers = 0x1;
+			}
 			tr->langopts.param[LOPT_WORD_MERGE] = 1;   // don't break vowels betwen words
 			SetIndicLetters(tr);   // call this after setting OFFSET_
 		}

src/translate.cpp

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2005 to 2010 by Jonathan Duddington                     *
+ *   Copyright (C) 2005 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -395,6 +395,9 @@ int IsAlpha(unsigned int c)
 	if((c >= 0x300) && (c <= 0x36f))
 		return(1);   // combining accents
 
+	if((c >= 0x780) && (c <= 0x7b1))
+		return(1);   // taani/divehi (maldives)
+
 	if((c >= 0x1100) && (c <= 0x11ff))
 		return(1);  //Korean jamo
 
@@ -760,10 +763,11 @@ int ChangeEquivalentPhonemes(Translator *tr, int lang2, char *phonemes)
 
 
 
-int TranslateWord(Translator *tr, char *word1, int next_pause, WORD_TAB *wtab)
+int TranslateWord(Translator *tr, char *word_start, int next_pause, WORD_TAB *wtab)
 {//===========================================================================
 // word1 is terminated by space (0x20) character
 
+	char *word1;
 	int word_length;
 	int ix;
 	char *p;
@@ -781,7 +785,9 @@ int TranslateWord(Translator *tr, char *word1, int next_pause, WORD_TAB *wtab)
 	char unpron_phonemes[N_WORD_PHONEMES];
 	char end_phonemes[N_WORD_PHONEMES];
 	char word_copy[N_WORD_BYTES];
-	char prefix_chars[41];
+	char word_copy2[N_WORD_BYTES];
+	int word_copy_length;
+	char prefix_chars[0x3f + 2];  
 	int found=0;
    int end_flags;
 	char c_temp;   // save a character byte while we temporarily replace it with space
@@ -795,6 +801,7 @@ int TranslateWord(Translator *tr, char *word1, int next_pause, WORD_TAB *wtab)
 	int emphasize_allcaps = 0;
 	int wflags;
 	int wmark;
+	int was_unpronouncable = 0;
 	WORD_TAB wtab_null[8];
 
 	// translate these to get pronunciations of plural 's' suffix (different forms depending on
@@ -831,6 +838,7 @@ int TranslateWord(Translator *tr, char *word1, int next_pause, WORD_TAB *wtab)
 	}
 
 	// count the length of the word
+	word1 = word_start;
 	if(*word1 == ' ') word1++;   // possibly a dot was replaced by space:  $dot
 	wordx = word1;
 
@@ -842,6 +850,9 @@ int TranslateWord(Translator *tr, char *word1, int next_pause, WORD_TAB *wtab)
 		word_length++;
 	}
 
+	word_copy_length = wordx - word_start;
+	memcpy(word_copy2, word_start, word_copy_length);
+
 	spell_word = 0;
 
 	if(option_sayas == SAYAS_KEY)
@@ -958,8 +969,12 @@ if((wmark > 0) && (wmark < 8))
 
 			if((tr->langopts.numbers & NUM_ROMAN) || ((tr->langopts.numbers & NUM_ROMAN_CAPITALS) && (wflags & FLAG_ALL_UPPER)))
 			{
-				if((found = TranslateRoman(tr, word1, phonemes, wtab)) != 0)
-					dictionary_flags[0] |= FLAG_ABBREV;   // prevent emphasis if capitals
+				if((wflags & FLAG_LAST_WORD) || !(wtab[1].flags & FLAG_NOSPACE))
+				{
+					// don't use Roman number if this word is not separated from the next word (eg. "XLTest")
+					if((found = TranslateRoman(tr, word1, phonemes, wtab)) != 0)
+						dictionary_flags[0] |= FLAG_ABBREV;   // prevent emphasis if capitals
+				}
 			}
 		}
 
@@ -1011,6 +1026,7 @@ if((wmark > 0) && (wmark < 8))
 		{
 			// This word looks "unpronouncable", so speak letters individually until we
 			// find a remainder that we can pronounce.
+			was_unpronouncable = FLAG_WAS_UNPRONOUNCABLE;
 			emphasize_allcaps = 0;
 
 			if(wordx[0] == '\'')
@@ -1084,10 +1100,10 @@ if((wmark > 0) && (wmark < 8))
 
 			found = 0;
 			confirm_prefix = 1;
-			while(end_type & SUFX_P)
+			for (int loopcount = 0; (loopcount < 50) && (end_type & SUFX_P); loopcount++)
 			{
 				// Found a standard prefix, remove it and retranslate
-
+				// loopcount guards against an endless loop
 				if(confirm_prefix && !(end_type & SUFX_B))
 				{
 					int end2;
@@ -1484,6 +1500,8 @@ if(dictionary_flags2[0] & FLAG_ABBREV)
 		ApplySpecialAttribute2(tr,word_phonemes,dictionary_flags[0]);
 	}
 
+	dictionary_flags[0] |= was_unpronouncable;
+	memcpy(word_start, word_copy2, word_copy_length);
 	return(dictionary_flags[0]);
 }  //  end of TranslateWord
 
@@ -1709,10 +1727,11 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 		{
 			char *p2;
 			int ok = 1;
-			int flags2 = 0;
+			unsigned int flags2[2];
 			int c_word2;
 			char ph_buf[N_WORD_PHONEMES];
 
+			flags2[0] = 0;
 			sylimit = tr->langopts.param[LOPT_COMBINE_WORDS];
 
 			// LANG=cs,sk
@@ -1729,12 +1748,15 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 			if(ok != 0)
 			{
 				strcpy(ph_buf,word_phonemes);
-				flags2 = TranslateWord(translator, p2+1, 0, wtab+1);
+
+				flags2[0] = TranslateWord(translator, p2+1, 0, wtab+1);
+				if(flags2[0] & FLAG_WAS_UNPRONOUNCABLE)
+					ok = 0;
 
 				if(sylimit & 0x100)
 				{
 					// only if the second word has $alt attribute
-					if((flags2 & FLAG_ALT_TRANS) == 0)
+					if((flags2[0] & FLAG_ALT_TRANS) == 0)
 					{
 						ok = 0;
 					}
@@ -1765,7 +1787,7 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 				else
 				{
 					if(flags == 0)
-						flags = flags2;   // no flags for the combined word, so use flags from the second word eg. lang-hu "nem december 7-e"
+						flags = flags2[0];   // no flags for the combined word, so use flags from the second word eg. lang-hu "nem december 7-e"
 					flags |= FLAG_SKIPWORDS;
 					dictionary_skipwords = 1;
 				}
@@ -2288,6 +2310,7 @@ void *TranslateClause(Translator *tr, FILE *f_text, const void *vp_input, int *t
 	int cc;
 	unsigned int source_index=0;
 	unsigned int prev_source_index=0;
+	int source_index_word=0;
 	int prev_in;
 	int prev_out=' ';
 	int prev_out2;
@@ -2329,6 +2352,11 @@ void *TranslateClause(Translator *tr, FILE *f_text, const void *vp_input, int *t
 	int tone;
 	int tone2;
 
+	if(tr==NULL)
+	{
+		return(NULL);
+	}
+
 	p_textinput = (unsigned char *)vp_input;
 	p_wchar_input = (wchar_t *)vp_input;
 
@@ -2431,7 +2459,7 @@ p = source;
 	words[0].flags = 0;
 	finished = 0;
 
-	for(j=0; charix[j]==0; j++);
+	for(j=0; charix[j]<=0; j++);
 	words[0].sourceix = charix[j];
 	k = 0;
 	while(charix[j] != 0)
@@ -2774,6 +2802,13 @@ if((c == '/') && (tr->langopts.testing & 2) && IsDigit09(next_in) && IsAlpha(pre
 			else
 			if(c == '.')
 			{
+				if(prev_out == '.')
+				{
+					// multiple dots, separate by spaces. Note >3 dots has been replaced by elipsis
+					c = ' ';
+					space_inserted = 1;
+				}
+				else
 				if(!(words[word_count-1].flags & FLAG_NOSPACE) && IsAlpha(prev_in))
 				{
 					// dot after a word, with space following, probably an abbreviation
@@ -2878,9 +2913,20 @@ if((c == '/') && (tr->langopts.testing & 2) && IsDigit09(next_in) && IsAlpha(pre
 
 			if(space_inserted)
 			{
-				words[word_count].length = source_index - words[word_count].sourceix;
+				// count the number of characters since the start of the word
+				j = 0;
+				k = source_index - 1;
+				while((k >= source_index_word) && (charix[k] != 0))
+				{
+					if(charix[k] > 0)    // don't count initial bytes of multi-byte character
+						j++;
+					k--;
+				}
+				words[word_count].length = j;
 			}
 
+			source_index_word = source_index;
+
 			// end of 'word'
 			sbuf[ix++] = ' ';
 

src/translate.h

@@ -1,5 +1,5 @@
 /***************************************************************************
- *   Copyright (C) 2005 to 2010 by Jonathan Duddington                     *
+ *   Copyright (C) 2005 to 2011 by Jonathan Duddington                     *
  *   email: jonsd@users.sourceforge.net                                    *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -60,7 +60,7 @@
 
 #define FLAG_ALLOW_DOT  0x01000000  // ignore '.' after word (abbreviation)
 #define FLAG_NEEDS_DOT  0x02000000  // only if the word is followed by a dot
-
+#define FLAG_WAS_UNPRONOUNCABLE  0x04000000  // the unpronounceable routine was used
 #define FLAG_MAX3       0x08000000  // limit to 3 repeats
 #define FLAG_PAUSE1     0x10000000  // shorter prepause
 #define FLAG_TEXTMODE   0x20000000  // word translates to replacement text, not phonemes
@@ -485,10 +485,6 @@ typedef struct {
 
 
 
-#define NUM_SEP_DOT    0x0008    // . , for thousands and decimal separator
-#define NUM_SEP_SPACE  0x1000    // allow space as thousands separator (in addition to langopts.thousands_sep)
-#define NUM_DEC_IT     0x2000    // (LANG=it) speak post-decimal-point digits as a combined number not as single digits
-
 typedef struct
 {//===========
 

src/transldlg.cpp

@@ -130,10 +130,15 @@ void PlayWavFile(const char *fname)
 #ifdef PLATFORM_WINDOWS
 	wxSound(wxString(fname,wxConvLocal)).Play(wxSOUND_SYNC);
 #else
-	sprintf(command,"play %s",fname);
+//	wxSound(wxString(fname,wxConvLocal)).Play(wxSOUND_SYNC);  // This gives a click at the start of each play
+	sprintf(command,"aplay %s",fname);
 	if(system(command) == -1)
 	{
-		wxLogError(_T("Failed to run system command:\n\n"+wxString(command,wxConvLocal)));
+		sprintf(command,"play %s",fname);
+		if(system(command) == -1)
+		{
+			wxLogError(_T("Failed to run system command:\n\n"+wxString(command,wxConvLocal)));
+		}
 	}
 #endif
 }
@@ -409,6 +414,12 @@ void TranslDlg::OnCommand(wxCommandEvent& event)
 	static int n_ph_list;
 	static PHONEME_LIST ph_list[N_PH_LIST+1];
 
+	if(translator==NULL)
+	{
+		wxLogError(_T("Voice not set"));
+		return;
+	}
+
 	option_phonemes = 0;
 
 	switch(event.GetId())

src/voices.cpp

@@ -411,7 +411,7 @@ void VoiceReset(int tone_only)
 	voice->samplerate = 22050;
 	memset(voice->klattv,0,sizeof(voice->klattv));
 
-	speed.fast_settings[0] = 350;
+	speed.fast_settings[0] = 400;
 	speed.fast_settings[1] = 800;
 	speed.fast_settings[2] = 175;
 
@@ -1506,6 +1506,8 @@ char const *SelectVoice(espeak_VOICE *voice_select, int *found)
 	}
 
 	// index the sorted list by the required variant number
+	if(ix2 == 0)
+		return(NULL);
 	vp = voices2[voice_select2.variant % ix2];
 
 	if(vp->variant != 0)

src/wave_sada.cpp

@@ -63,6 +63,7 @@ static uint32_t total_samples_skipped;
 //
 static uint32_t last_play_position=0;
 
+static uint32_t wave_samplerate;
 //>
 // wave_init 
 //

src/wavegen.cpp

@@ -1965,6 +1965,10 @@ int WavegenFill2(int fill_zeros)
 			if((wdata.amplitude_fmt = q[1]) == 0)
 				wdata.amplitude_fmt = 100;  // percentage, but value=0 means 100%
 			break;
+
+		case WCMD_SONIC_SPEED:
+			sonicSpeed = (double)q[1] / 1024;
+			break;
 		}
 
 		if(result==0)
@@ -1999,6 +2003,9 @@ static int SpeedUp(short *outbuf, int length_in, int length_out, int end_of_text
 		sonicWriteShortToStream(sonicSpeedupStream, outbuf, length_in);
 	}
 
+	if(sonicSpeedupStream == NULL)
+		return(0);
+
 	if(end_of_text)
 	{
 		sonicFlushStream(sonicSpeedupStream);