Merge branch 'master' into android

Conflicts:
	platforms/windows/make_espeakedit.iss

src/Makefile.am

@@ -6,7 +6,7 @@ DATADIR=${PREFIX}/share/espeak-data
 
 # remove -fno-exceptions  if it causes problems with stdio library on some platforms (ARM)
 
-RELEASE = 1.46
+RELEASE = 1.47
 SPEAK = speak
 ESPEAK = espeak
 ESPEAKEDIT = espeakedit

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

@@ -611,7 +611,7 @@ 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))
@@ -1499,8 +1499,8 @@ void SetWordStress(Translator *tr, char *output, unsigned int *dictionary_flags,
 			else
 			{
 				// no long vowels or consonant clusters
-				if(vowel_count > 4)
-					stressed_syllable = vowel_count - 3;
+				if(vowel_count > 5)
+					stressed_syllable = vowel_count - 3;  // more than 4 syllables
 				else
 					stressed_syllable = vowel_count - 1;
 			}
@@ -2698,7 +2698,7 @@ int TranslateRules(Translator *tr, char *p_start, char *phonemes, int ph_size, c
 						{
 							if(tr->langopts.alt_alphabet == alphabet->offset)
 							{
-								sprintf(phonemes,"%c%s",phonSWITCH, tr->langopts.alt_alphabet_lang);
+								sprintf(phonemes,"%c%s",phonSWITCH, WordToString2(tr->langopts.alt_alphabet_lang));
 								return(0);
 							}
 							if(alphabet->flags & AL_WORDS)
@@ -2988,6 +2988,7 @@ static const char *LookupDict2(Translator *tr, const char *word, const char *wor
 	const char *word_end;
 	const char *word1;
 	int wflags = 0;
+	int lookup_symbol;
 	char word_buf[N_WORD_BYTES+1];
 	char dict_flags_buf[80];
 
@@ -2996,6 +2997,7 @@ static const char *LookupDict2(Translator *tr, const char *word, const char *wor
 		wflags = wtab->flags;
 	}
 
+	lookup_symbol = flags[1] & FLAG_LOOKUP_SYMBOL;
 	word1 = word;
 	if(tr->transpose_min > 0)
 	{
@@ -3182,9 +3184,9 @@ static const char *LookupDict2(Translator *tr, const char *word, const char *wor
 				continue;
 		}
 
-		if((dictionary_flags2 & FLAG_ATEND) && (word_end < translator->clause_end))
+		if((dictionary_flags2 & FLAG_ATEND) && (word_end < translator->clause_end) && (lookup_symbol==0))
 		{
-			// only use this pronunciation if it's the last word of the clause
+			// only use this pronunciation if it's the last word of the clause, or called from Lookup()
 			continue;
 		}
 
@@ -3196,7 +3198,7 @@ static const char *LookupDict2(Translator *tr, const char *word, const char *wor
 
 		if((dictionary_flags2 & FLAG_SENTENCE) && !(translator->clause_terminator & CLAUSE_BIT_SENTENCE))
 		{
-			// only uis this clause is a sentence , i.e. terminator is {. ? !} not {, : :}
+			// only if this clause is a sentence , i.e. terminator is {. ? !} not {, : :}
 			continue;
 		}
 
@@ -3364,7 +3366,7 @@ 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;
@@ -3475,7 +3477,8 @@ int LookupDictList(Translator *tr, char **wordptr, char *ph_out, unsigned int *f
 int Lookup(Translator *tr, const char *word, char *ph_out)
 {//===================================================
 	unsigned int flags[2];
-	flags[0] = flags[1] = 0;
+	flags[0] = 0;
+	flags[1] = FLAG_LOOKUP_SYMBOL;
 	char *word1 = (char *)word;
 	return(LookupDictList(tr, &word1, ph_out, flags, 0, NULL));
 }

src/espeak.cpp

@@ -128,13 +128,16 @@ void DisplayVoices(FILE *f_out, char *language)
 	const char *p;
 	int len;
 	int count;
+	int c;
+	int j;
 	const espeak_VOICE *v;
 	const char *lang_name;
 	char age_buf[12];
+	char buf[80];
 	const espeak_VOICE **voices;
 	espeak_VOICE voice_select;
 
-	static char genders[4] = {' ','M','F',' '};
+	static char genders[4] = {'-','M','F','-'};
 
 	if((language != NULL) && (language[0] != 0))
 	{
@@ -150,7 +153,7 @@ void DisplayVoices(FILE *f_out, char *language)
 		voices = espeak_ListVoices(NULL);
 	}
 
-	fprintf(f_out,"Pty Language Age/Gender VoiceName       File        Other Langs\n");
+	fprintf(f_out,"Pty Language Age/Gender VoiceName          File          Other Languages\n");
 
 	for(ix=0; (v = voices[ix]) != NULL; ix++)
 	{
@@ -168,8 +171,16 @@ void DisplayVoices(FILE *f_out, char *language)
 
 			if(count==0)
 			{
-				fprintf(f_out,"%2d  %-12s%s%c  %-17s %-11s ",
-               p[0],lang_name,age_buf,genders[v->gender],v->name,v->identifier);
+				for(j=0; j < sizeof(buf); j++)
+				{
+					// replace spaces in the name
+					if((c = v->name[j]) == ' ')
+						c = '_';
+					if((buf[j] = c) == 0)
+						break;
+				}
+				fprintf(f_out,"%2d  %-12s%s%c  %-20s %-13s ",
+               p[0],lang_name,age_buf,genders[v->gender],buf,v->identifier);
 			}
 			else
 			{

src/numbers.cpp

@@ -608,7 +608,7 @@ 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;
@@ -695,6 +695,9 @@ int TranslateLetter(Translator *tr, char *word, char *phonemes, int control)
 
 	if(ph_buf[0] == 0)
 	{
+		if((al_offset != 0) && (al_offset == translator->langopts.alt_alphabet))
+			language = translator->langopts.alt_alphabet_lang;
+		else
 		if((alphabet != NULL) && (alphabet->language != 0) && !(al_flags & AL_NOT_LETTERS))
 			language = alphabet->language;
 		else
@@ -797,7 +800,7 @@ int TranslateLetter(Translator *tr, char *word, char *phonemes, int control)
 
 	len = strlen(phonemes);
 
-	if(tr->langopts.accents & 2)
+	if(tr->langopts.accents & 2)  // 'capital' before or after the word ?
 		sprintf(ph_buf2,"%c%s%s%s",0xff,ph_alphabet,ph_buf,capital);
 	else
 		sprintf(ph_buf2,"%c%s%s%s",0xff,ph_alphabet,capital,ph_buf);  // the 0xff marker will be removed or replaced in SetSpellingStress()
@@ -877,6 +880,7 @@ void SetSpellingStress(Translator *tr, char *phonemes, int control, int n_chars)
 // Numbers
 
 static char ph_ordinal2[12];
+static char ph_ordinal2x[12];
 
 
 static int CheckDotOrdinal(Translator *tr, char *word, char *word_end, WORD_TAB *wtab, int roman)
@@ -981,7 +985,7 @@ 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 +1024,7 @@ 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;
@@ -1303,7 +1307,11 @@ static int LookupNum2(Translator *tr, int value, const int control, char *ph_out
 				if(control & 4)
 				{
 					sprintf(string,"_%d%cx",value,ord_type);  // LANG=hu, special word for 1. 2. when there are no higher digits
-					found = Lookup(tr, string, ph_digits);
+					if((found = Lookup(tr, string, ph_digits)) != 0)
+					{
+						if(ph_ordinal2x[0] != 0)
+							strcpy(ph_ordinal, ph_ordinal2x);  // alternate pronunciation (lang=an)
+					}
 				}
 				if(found == 0)
 				{
@@ -1363,8 +1371,7 @@ static int LookupNum2(Translator *tr, int value, const int control, char *ph_out
 			else
 			{
 
-				if((is_ordinal) &&
-						((units == 0) || (tr->langopts.numbers & NUM_SWAP_TENS) || (tr->langopts.numbers2 & NUM2_MULTIPLE_ORDINAL)))
+				if(is_ordinal)
 				{
 					sprintf(string,"_%dX%c", tens, ord_type);
 					if(Lookup(tr, string, ph_tens) != 0)
@@ -1458,7 +1465,7 @@ static int LookupNum2(Translator *tr, int value, const int control, char *ph_out
 		{
 			Lookup(tr, "_0and", ph_and);
 
-			if((is_ordinal) && (tr->langopts.numbers2 & NUM2_MULTIPLE_ORDINAL))
+			if((is_ordinal) && (tr->langopts.numbers2 & NUM2_ORDINAL_NO_AND))
 				ph_and[0] = 0;
 
 			if(tr->langopts.numbers & NUM_SWAP_TENS)
@@ -1607,9 +1614,9 @@ static int LookupNum3(Translator *tr, int value, char *ph_out, int suppress_null
 			}
 
 			if(tr->langopts.numbers2 & 0x200)
-				sprintf(ph_thousands,"%s%s",ph_10T,ph_digits);  // say "thousands" before its number, not after
+				sprintf(ph_thousands,"%s%c%s%c",ph_10T,phonEND_WORD,ph_digits,phonEND_WORD);  // say "thousands" before its number, not after
 			else
-				sprintf(ph_thousands,"%s%s",ph_digits,ph_10T);
+				sprintf(ph_thousands,"%s%c%s%c",ph_digits,phonEND_WORD,ph_10T,phonEND_WORD);
 
 			hundreds %= 10;
 			if((hundreds == 0) && (say_zero_hundred == 0))
@@ -1742,7 +1749,7 @@ static int LookupNum3(Translator *tr, int value, char *ph_out, int suppress_null
 		}
 	}
 
-	sprintf(ph_out,"%s%s%s",buf1,ph_hundred_and,buf2);
+	sprintf(ph_out,"%s%s%c%s",buf1,ph_hundred_and,phonEND_WORD,buf2);
 
 	return(0);
 }  // end of LookupNum3
@@ -1753,12 +1760,12 @@ 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);
@@ -1790,6 +1797,7 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 	int suffix_ix;
 	int skipwords = 0;
 	int group_len;
+	int len;
 	char *p;
 	char string[32];  // for looking up entries in **_list
 	char buf1[100];
@@ -1808,7 +1816,7 @@ 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);
 
@@ -1817,14 +1825,14 @@ 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;
 		}
@@ -1846,7 +1854,7 @@ 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;
@@ -1875,7 +1883,7 @@ 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))
@@ -1884,6 +1892,8 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 					ordinal = 2;
 					flags[0] |= FLAG_SKIPWORDS;
 					skipwords = 1;
+					sprintf(string,"_x#%s",suffix);
+					Lookup(tr, string, ph_ordinal2x);  // is there an alternate pronunciation?
 				}
 			}
 		}
@@ -1898,7 +1908,7 @@ 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
 		}
@@ -1967,7 +1977,7 @@ 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);
@@ -2012,8 +2022,8 @@ 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
@@ -2066,9 +2076,9 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 
 	LookupNum3(tr, value, ph_buf, suppress_null, thousandplex, prev_thousands | ordinal | decimal_point);
 	if((thousandplex > 0) && (tr->langopts.numbers2 & 0x200))
-		sprintf(ph_out,"%s%s%s%s",ph_zeros,ph_append,ph_buf2,ph_buf);  // say "thousands" before its number
+		sprintf(ph_out,"%s%s%c%s%s",ph_zeros,ph_append,phonEND_WORD,ph_buf2,ph_buf);  // say "thousands" before its number
 	else
-		sprintf(ph_out,"%s%s%s%s",ph_zeros,ph_buf2,ph_buf,ph_append);
+		sprintf(ph_out,"%s%s%s%c%s",ph_zeros,ph_buf2,ph_buf,phonEND_WORD,ph_append);
 
 
 	while(decimal_point)
@@ -2076,7 +2086,7 @@ 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)
@@ -2095,7 +2105,7 @@ 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);
@@ -2146,19 +2156,20 @@ 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';
 			LookupNum2(tr, value, 2, buf1);
-			strcat(ph_out,buf1);
+			len = strlen(ph_out);
+			sprintf(&ph_out[len],"%c%s", phonEND_WORD, buf1);
 		}
 
 		// something after the decimal part ?
 		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);
@@ -2178,7 +2189,8 @@ static int TranslateNumber_1(Translator *tr, char *word, char *ph_out, unsigned
 		if((tr->langopts.numbers & NUM_NOPAUSE) && (next_char == ' '))
 			utf8_in(&next_char,p);
 
-		if(!iswalpha(next_char) && !((wtab[thousandplex].flags & FLAG_HYPHEN_AFTER) && (thousands_exact != 0)))
+		if(!iswalpha(next_char) && (thousands_exact==0))
+//		if(!iswalpha(next_char) && !((wtab[thousandplex].flags & FLAG_HYPHEN_AFTER) && (thousands_exact != 0)))
 			strcat(ph_out,str_pause);  // don't add pause for 100s,  6th, etc.
 	}
 

src/phonemelist.cpp

@@ -124,6 +124,7 @@ void MakePhonemeList(Translator *tr, int post_pause, int start_sentence)
 	int alternative;
 	int delete_count;
 	int word_start;
+	int inserted;
 	PHONEME_DATA phdata;
 
 	int n_ph_list3;
@@ -341,9 +342,7 @@ void MakePhonemeList(Translator *tr, int post_pause, int start_sentence)
 	{
 		plist3 = &ph_list3[j];
 
-		if(plist3->sourceix != 0)
-			word_start = j;
-
+		inserted = 0;
 		if(insert_ph != 0)
 		{
 			// we have a (linking) phoneme which we need to insert here
@@ -374,10 +373,14 @@ void MakePhonemeList(Translator *tr, int post_pause, int start_sentence)
 			ph = phoneme_tab[insert_ph];
 			plist3->ph = ph;
 			insert_ph = 0;
+			inserted = 1;    // don't insert the same phoneme repeatedly
 		}
 		else
 		{
 			// otherwise get the next phoneme from the list
+			if(plist3->sourceix != 0)
+				word_start = j;
+
 			ph = phoneme_tab[plist3->phcode];
 			plist3[0].ph = ph;
 
@@ -394,7 +397,7 @@ void MakePhonemeList(Translator *tr, int post_pause, int start_sentence)
 
 		InterpretPhoneme(tr, 0x100, plist3, &phdata, &worddata);
 
-		if((alternative = phdata.pd_param[pd_INSERTPHONEME]) > 0)
+		if(((alternative = phdata.pd_param[pd_INSERTPHONEME]) > 0) && (inserted == 0))
 		{
 			// PROBLEM: if we insert a phoneme before a vowel then we loose the stress.
 			PHONEME_TAB *ph2;

src/readclause.cpp

@@ -1292,10 +1292,10 @@ 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/speak.cpp

@@ -160,13 +160,16 @@ void DisplayVoices(FILE *f_out, char *language)
 	const char *p;
 	int len;
 	int count;
+	int c;
+	int j;
 	const espeak_VOICE *v;
 	const char *lang_name;
 	char age_buf[12];
+	char buf[80];
 	const espeak_VOICE **voices;
 	espeak_VOICE voice_select;
 
-	static char genders[4] = {' ','M','F',' '};
+	static char genders[4] = {'-','M','F','-'};
 
 	if((language != NULL) && (language[0] != 0))
 	{
@@ -182,7 +185,7 @@ void DisplayVoices(FILE *f_out, char *language)
 		voices = espeak_ListVoices(NULL);
 	}
 
-	fprintf(f_out,"Pty Language Age/Gender VoiceName       File        Other Langs\n");
+	fprintf(f_out,"Pty Language Age/Gender VoiceName          File          Other Languages\n");
 
 	for(ix=0; (v = voices[ix]) != NULL; ix++)
 	{
@@ -200,8 +203,16 @@ void DisplayVoices(FILE *f_out, char *language)
 
 			if(count==0)
 			{
-				fprintf(f_out,"%2d  %-12s%s%c  %-17s %-11s ",
-               p[0],lang_name,age_buf,genders[v->gender],v->name,v->identifier);
+				for(j=0; j < sizeof(buf); j++)
+				{
+					// replace spaces in the name
+					if((c = v->name[j]) == ' ')
+						c = '_';
+					if((buf[j] = c) == 0)
+						break;
+				}
+				fprintf(f_out,"%2d  %-12s%s%c  %-20s %-13s ",
+               p[0],lang_name,age_buf,genders[v->gender],buf,v->identifier);
 			}
 			else
 			{
@@ -750,11 +761,22 @@ int main (int argc, char **argv)
 	init_path(argv[0],data_path);
 	initialise();
 
+	if(voicename[0] == 0)
+		strcpy(voicename,"default");
+
+	if(SetVoiceByName(voicename) != EE_OK)
+	{
+		memset(&voice_select,0,sizeof(voice_select));
+		voice_select.languages = voicename;
+		if(SetVoiceByProperties(&voice_select) != EE_OK)
+		{
+			fprintf(stderr,"%svoice '%s'\n",err_load,voicename);
+			exit(2);
+		}
+	}
 
 	if(flag_compile)
 	{
-		LoadVoice(voicename,5);
-
 #ifdef PLATFORM_DOS
 		char path_dsource[sizeof(path_home)+20];
 		strcpy(path_dsource,path_home);
@@ -776,20 +798,6 @@ int main (int argc, char **argv)
 	}
 
 
-	if(voicename[0] == 0)
-		strcpy(voicename,"default");
-
-	if(SetVoiceByName(voicename) != EE_OK)
-	{
-		memset(&voice_select,0,sizeof(voice_select));
-		voice_select.languages = voicename;
-		if(SetVoiceByProperties(&voice_select) != EE_OK)
-		{
-			fprintf(stderr,"%svoice '%s'\n",err_load,voicename);
-			exit(2);
-		}
-	}
-
 	SetParameter(espeakRATE,speed,0);
 	SetParameter(espeakVOLUME,amp,0);
 	SetParameter(espeakCAPITALS,option_capitals,0);

src/synthdata.cpp

@@ -35,8 +35,8 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.46.53  16.Mar.13";
-const int version_phdata  = 0x014640;
+const char *version_string = "1.47.03c  23.Mar.13";
+const int version_phdata  = 0x014701;
 
 int option_device_number = -1;
 FILE *f_logespeak = NULL;
@@ -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

@@ -103,6 +103,19 @@ ALPHABET alphabets [] = {
 };
 
 
+ALPHABET *AlphabetFromName(const char *name)
+{//==========================================
+	ALPHABET *alphabet;
+
+	for(alphabet=alphabets; alphabet->name != NULL; alphabet++)
+	{
+		if(strcmp(name, &alphabet->name[1]) == 0)
+			return(alphabet);
+	}
+	return(NULL);
+}
+
+
 ALPHABET *AlphabetFromChar(int c)
 {//===============================
 	// Find the alphabet from a character.
@@ -276,8 +289,8 @@ static const char transpose_map_latin[] = {
 	tr->langopts.param2[LOPT_BRACKET_PAUSE] = 2;    // pauses when announcing bracket names
 	tr->langopts.max_initial_consonants = 3;
 	tr->langopts.replace_chars = NULL;
-	tr->langopts.ascii_language = "";    // Non-Latin alphabet languages, use this language to speak Latin words, default is English
-
+	tr->langopts.ascii_language[0] = 0;    // Non-Latin alphabet languages, use this language to speak Latin words, default is English
+	tr->langopts.alt_alphabet_lang = L('e','n');
 
 	SetLengthMods(tr,201);
 //	tr->langopts.length_mods = length_mods_en;
@@ -526,9 +539,10 @@ Translator *SelectTranslator(const char *name)
 			tr->langopts.stress_flags =  S_MID_DIM | S_FINAL_DIM;   // use 'diminished' for unstressed final syllable
 			tr->letter_bits_offset = OFFSET_BENGALI;
 			SetIndicLetters(tr);   // call this after setting OFFSET_BENGALI
+			SetLetterBitsRange(tr,LETTERGP_B,0x01,0x01);   // candranindu
 			SetLetterBitsRange(tr,LETTERGP_F,0x3e,0x4c);   // vowel signs, but not virama
 
-			tr->langopts.numbers = 0x1;
+			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
 		}
 		break;
@@ -666,7 +680,7 @@ Translator *SelectTranslator(const char *name)
 			tr->langopts.param[LOPT_SONORANT_MIN] = 130;  // limit the shortening of sonorants before short vowels
 
 			tr->langopts.numbers = NUM_SINGLE_STRESS | NUM_DECIMAL_COMMA;
-			tr->langopts.numbers2 = 0x2 | NUM2_MULTIPLE_ORDINAL;   // variant form of numbers before thousands
+			tr->langopts.numbers2 = 0x2 | NUM2_MULTIPLE_ORDINAL | NUM2_ORDINAL_NO_AND;   // variant form of numbers before thousands
 
 			if(name2 == L_grc)
 			{
@@ -720,7 +734,7 @@ Translator *SelectTranslator(const char *name)
 			tr->langopts.param[LOPT_SONORANT_MIN] = 120;  // limit the shortening of sonorants before short vowels
 
 			tr->langopts.numbers = NUM_SINGLE_STRESS | NUM_DECIMAL_COMMA | NUM_AND_UNITS | NUM_OMIT_1_HUNDRED | NUM_OMIT_1_THOUSAND | NUM_ROMAN | NUM_ROMAN_AFTER;
-			tr->langopts.numbers2 = NUM2_MULTIPLE_ORDINAL;
+			tr->langopts.numbers2 = NUM2_MULTIPLE_ORDINAL | NUM2_ORDINAL_NO_AND;
 
 			if(name2 == L('c','a'))
 			{
@@ -732,7 +746,7 @@ Translator *SelectTranslator(const char *name)
 			if(name2 == L('a','n'))
 			{
 				tr->langopts.stress_flags = S_FINAL_SPANISH | S_FINAL_DIM_ONLY | S_FINAL_NO_2;
-				tr->langopts.numbers2 = 0;
+				tr->langopts.numbers2 = NUM2_ORDINAL_NO_AND;
 			}
 			else
 			if(name2 == L_pap)
@@ -828,6 +842,7 @@ Translator *SelectTranslator(const char *name)
 		{
 			tr->langopts.stress_rule = STRESSPOSN_1L;
 			tr->langopts.numbers = 1;
+			tr->langopts.accents = 2;  // 'capital' after letter name
 		}
 		break;
 
@@ -960,6 +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 };
@@ -1054,7 +1070,7 @@ SetLengthMods(tr,3);  // all equal
 			tr->langopts.numbers = NUM_VIGESIMAL | NUM_AND_UNITS | NUM_OMIT_1_HUNDRED |NUM_OMIT_1_THOUSAND | NUM_DFRACTION_5 | NUM_ROMAN;
 
 			tr->langopts.alt_alphabet = OFFSET_CYRILLIC;
-			tr->langopts.alt_alphabet_lang = "ru";
+			tr->langopts.alt_alphabet_lang = L('r','u');
 		}
 		break;
 
@@ -1084,6 +1100,7 @@ SetLengthMods(tr,3);  // all equal
 
 	case L('k','l'):   // Greenlandic
 		{
+			SetupTranslator(tr,stress_lengths_equal,stress_amps_equal);
 			tr->langopts.stress_rule = 12;
 			tr->langopts.stress_flags = S_NO_AUTO_2;
 			tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SWAP_TENS | NUM_HUNDRED_AND | NUM_OMIT_1_HUNDRED | NUM_ORDINAL_DOT | NUM_1900 | NUM_ROMAN | NUM_ROMAN_CAPITALS | NUM_ROMAN_ORDINAL;
@@ -1263,7 +1280,7 @@ SetLengthMods(tr,3);  // all equal
 			tr->langopts.stress_rule = STRESSPOSN_1R;        // stress on final syllable
 			tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2 | S_INITIAL_2 | S_PRIORITY_STRESS;
 			tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_DFRACTION_2 | NUM_HUNDRED_AND | NUM_AND_UNITS | NUM_ROMAN;
-			tr->langopts.numbers2 = NUM2_MULTIPLE_ORDINAL | NUM2_NO_TEEN_ORDINALS;
+			tr->langopts.numbers2 = NUM2_MULTIPLE_ORDINAL | NUM2_NO_TEEN_ORDINALS | NUM2_ORDINAL_NO_AND;
 			SetLetterVowel(tr,'y');
 			ResetLetterBits(tr,0x2);
 			SetLetterBits(tr,1,"bcdfgjkmnpqstvxz");      // B  hard consonants, excluding h,l,r,w,y
@@ -1296,7 +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
 		}
@@ -1413,6 +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

@@ -378,6 +378,7 @@ void SetLengthMods(Translator *tr, int value)
 }
 
 
+
 int IsAlpha(unsigned int c)
 {//========================
 // Replacement for iswalph() which also checks for some in-word symbols
@@ -433,8 +434,8 @@ 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);
@@ -971,7 +972,7 @@ int TranslateWord(Translator *tr, char *word_start, int next_pause, WORD_TAB *wt
 				}
 			}
 		}
-		else if((found==0) && (dictionary_flags[0] & FLAG_SKIPWORDS))
+		else if((found==0) && (dictionary_flags[0] & FLAG_SKIPWORDS) && !(dictionary_flags[0] & FLAG_ABBREV))
 		{
 			// grouped words, but no translation.  Join the words with hyphens.
 			wordx = word1;
@@ -2315,10 +2316,10 @@ 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++;
 	}
 
@@ -3335,7 +3336,7 @@ 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

@@ -83,6 +83,7 @@
 #define FLAG_STEM          0x10000  // must have a suffix
 #define FLAG_ATEND         0x20000  /* use this pronunciation if at end of clause */
 #define FLAG_ATSTART       0x40000  // use this pronunciation if at start of clause
+#define FLAG_LOOKUP_SYMBOL 0x40000000  // to indicate called from Lookup()
 
 #define BITNUM_FLAG_ALLCAPS   0x2a
 #define BITNUM_FLAG_HYPHENATED  0x2c
@@ -502,6 +503,7 @@ typedef struct {
 #define NUM2_THOUSANDS_VAR4     0x100
 #define NUM2_THOUSANDS_VAR5     0x140
 
+#define NUM2_ORDINAL_NO_AND     0x800
 #define NUM2_MULTIPLE_ORDINAL   0x1000
 #define NUM2_NO_TEEN_ORDINALS   0x2000
 #define NUM2_MYRIADS            0x4000
@@ -510,6 +512,7 @@ typedef struct {
 	// bits 1-4  use variant form of numbers before thousands,millions,etc.
 	// bits 6-8  use different forms of thousand, million, etc (M MA MB)
 	// bit9=(LANG=rw) say "thousand" and "million" before its number, not after
+	// bit11=(LANG=es,an) don't say 'and' between tens and units for ordinal numbers
 	// bit12=(LANG=el,es) use ordinal form of hundreds and tens as well as units
 	// bit13=(LANG=pt) don't use 11-19 numbers to make ordinals
 	// bit14=(LANG=ko)  use myriads (groups of 4 digits) not thousands (groups of 3)
@@ -542,9 +545,9 @@ typedef struct {
 	int testing;            // testing options: bit 1= specify stressed syllable in the form:  "outdoor/2"
 	int listx;    // compile *_listx after *list
 	const unsigned int *replace_chars;      // characters to be substitutes
-	const char *ascii_language;  // switch to this language for Latin characters
+	char ascii_language[8];  // switch to this language for Latin characters
 	int alt_alphabet;       // offset for another language to recognize
-	const char *alt_alphabet_lang;  // language for the alt_alphabet
+	int alt_alphabet_lang;  // language for the alt_alphabet
 	int max_lengthmod;
 	int lengthen_tonic;   // lengthen the tonic syllable
 	int suffix_add_e;      // replace a suffix (which has the SUFX_E flag) with this character
@@ -706,6 +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);
@@ -713,6 +717,7 @@ int towlower2(unsigned int c);
 void GetTranslatedPhonemeString(char *phon_out, int n_phon_out, int use_ipa);
 const char *WordToString2(unsigned int word);
 ALPHABET *AlphabetFromChar(int c);
+ALPHABET *AlphabetFromName(const char *name);
 
 Translator *SelectTranslator(const char *name);
 int SetTranslator2(const char *name);

src/voices.cpp

@@ -110,6 +110,7 @@ enum {
 	V_FAST,
 	V_SPEED,
 	V_DICTMIN,
+	V_ALPHABET2,
 
 // these need a phoneme table to have been specified
 	V_REPLACE,
@@ -161,6 +162,7 @@ static MNEM_TAB keyword_tab[] = {
 	{"fast_test2",  V_FAST},
 	{"speed",      V_SPEED},
 	{"dict_min",   V_DICTMIN},
+	{"alphabet2",  V_ALPHABET2},
 
 	// these just set a value in langopts.param[]
 	{"l_dieresis", 0x100+LOPT_DIERESES},
@@ -396,10 +398,10 @@ void VoiceReset(int tone_only)
 // Set voice to the default values
 
 	int  pk;
-//	static unsigned char default_heights[N_PEAKS] = {140,128,120,110,90,90,128,128,128};  // changed for v.1.47
-//	static unsigned char default_widths[N_PEAKS] = {128,128,128,160,171,171,128,128,128};
-	static unsigned char default_heights[N_PEAKS] = {128,128,120,120,110,110,128,128,128};  // previous version
+	static unsigned char default_heights[N_PEAKS] = {130,128,120,116,100,100,128,128,128};  // changed for v.1.47
 	static unsigned char default_widths[N_PEAKS] = {140,128,128,160,171,171,128,128,128};
+//	static unsigned char default_heights[N_PEAKS] = {128,128,120,120,110,110,128,128,128};  // previous version
+//	static unsigned char default_widths[N_PEAKS] = {128,128,128,160,171,171,128,128,128};
 
 	static int breath_widths[N_PEAKS] = {0,200,200,400,400,400,600,600,600};
 
@@ -420,7 +422,7 @@ void VoiceReset(int tone_only)
 	voice->n_harmonic_peaks = 5;
 	voice->peak_shape = 0;
 	voice->voicing = 64;
-	voice->consonant_amp = 85;  // change to 85 for v.1.47 was 100
+	voice->consonant_amp = 90;  // change to 85 for v.1.47 was 100
 	voice->consonant_ampv = 100;
 	voice->samplerate = samplerate_native;
 	memset(voice->klattv,0,sizeof(voice->klattv));
@@ -448,7 +450,6 @@ void VoiceReset(int tone_only)
 		// adjust formant smoothing depending on sample rate
 		formant_rate[pk] = (formant_rate_22050[pk] * 22050)/samplerate;
 	}
-	voice->height[2] = 240;  // reduce F2 slightly
 
 	// This table provides the opportunity for tone control.
 	// Adjustment of harmonic amplitudes, steps of 8Hz
@@ -532,6 +533,21 @@ static int Read8Numbers(char *data_in,int *data)
 }
 
 
+static unsigned int StringToWord2(const char *string)
+{//======================================================
+// Convert a language name string to a word such as L('e','n')
+	int ix;
+	int c;
+	unsigned int value = 0;
+
+	for(ix=0; (ix<4) & ((c = string[ix]) != 0); ix++)
+	{
+		value = (value << 8) | (c & 0xff);
+	}
+	return(value);
+}
+
+
 voice_t *LoadVoice(const char *vname, int control)
 {//===============================================
 // control, bit 0  1= no_default
@@ -573,6 +589,7 @@ voice_t *LoadVoice(const char *vname, int control)
 	int stress_lengths[8];
 	int stress_add[8];
 	char names[8][40];
+	char name1[40];
 	char name2[80];
 	const char *voice_dir;
 
@@ -583,11 +600,11 @@ voice_t *LoadVoice(const char *vname, int control)
 	static char voice_name[40];        // voice name for current_voice_selected
 	static char voice_languages[100];  // list of languages and priorities for current_voice_selected
 
-	// which directory to look for a named voice
+	// 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 el es et fi fr-be hr hu is it lt lv mk nl no pl pt-pt ro ru sk sq sr sv ";
+		"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 ";
 
 
 	strncpy0(voicename, vname, sizeof(voicename));
@@ -962,12 +979,10 @@ voice_t *LoadVoice(const char *vname, int control)
 		case V_MBROLA:
 		{
 			int srate = 16000;
-			char name[40];
-			char phtrans[40];
 
-			phtrans[0] = 0;
-			sscanf(p,"%s %s %d",name,phtrans,&srate);
-			if(LoadMbrolaTable(name,phtrans,srate) != EE_OK)
+			name2[0] = 0;
+			sscanf(p,"%s %s %d",name1,name2,&srate);
+			if(LoadMbrolaTable(name1,name2,srate) != EE_OK)
 			{
 				fprintf(stderr,"mbrola voice not found\n");
 			}
@@ -990,6 +1005,28 @@ voice_t *LoadVoice(const char *vname, int control)
 			sscanf(p,"%d",&dict_min);
 			break;
 
+		case V_ALPHABET2:
+			{
+				ALPHABET *alphabet;
+				name1[0] = name2[0] = 0;
+				sscanf(p, "%s %s", name1, name2);
+
+				if(strcmp(name1, "latin") == 0)
+				{
+					strncpy0(langopts->ascii_language,name2,sizeof(langopts->ascii_language));
+				}
+				else if((alphabet = AlphabetFromName(name1)) != 0)
+				{
+					langopts->alt_alphabet = alphabet->offset;
+					langopts->alt_alphabet_lang = StringToWord2(name2);
+				}
+				else
+				{
+					fprintf(stderr,"alphabet name '%s' not found\n", name1);
+				}
+			}
+			break;
+
 		default:
 			if((key & 0xff00) == 0x100)
 			{
@@ -1098,7 +1135,7 @@ 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
 			}
@@ -1191,10 +1228,10 @@ static int ScoreVoice(espeak_VOICE *voice_spec, const char *spec_language, int s
 
 	p = voice->languages;  // list of languages+dialects for which this voice is suitable
 
-	if(strcmp(spec_language,"mbrola")==0)
+	if(spec_n_parts < 0)
 	{
-		// only list mbrola voices
-		if(memcmp(voice->identifier,"mb/",3) == 0)
+		// match on the subdirectory
+		if(memcmp(voice->identifier, spec_language, spec_lang_len) == 0)
 			return(100);
 		return(0);
 	}
@@ -1327,6 +1364,7 @@ static int SetVoiceScores(espeak_VOICE *voice_select, espeak_VOICE **voices, int
 	int lang_len=0;
 	espeak_VOICE *vp;
 	char language[80];
+	char buf[sizeof(path_home)+80];
 
 	// count number of parts in the specified language
 	if((voice_select->languages != NULL) && (voice_select->languages[0] != 0))
@@ -1339,6 +1377,26 @@ static int SetVoiceScores(espeak_VOICE *voice_select, espeak_VOICE **voices, int
 				n_parts++;
 		}
 	}
+
+	if((n_parts == 1) && (control & 1))
+	{
+		if(strcmp(language, "mbrola") == 0)
+		{
+			language[2] = 0;  // truncate to "mb"
+			lang_len = 2;
+		}
+
+		sprintf(buf, "%s/voices/%s", path_home, language);
+		if(GetFileLength(buf) == -2)
+		{
+			// A subdirectory name has been specified.  List all the voices in that subdirectory
+			language[lang_len++] = PATHSEP;
+			language[lang_len] = 0;
+			n_parts = -1;
+		}
+
+	}
+
 	// select those voices which match the specified language
 	nv = 0;
 	for(ix=0; ix<n_voices_list; ix++)
@@ -1866,11 +1924,12 @@ ESPEAK_API const espeak_VOICE **espeak_ListVoices(espeak_VOICE *voice_spec)
 	}
 	else
 	{
-		// list all: omit variant voices and mbrola voices
+		// list all: omit variant voices and mbrola voices and test voices
 		j = 0;
 		for(ix=0; (v = voices_list[ix]) != NULL; ix++)
 		{
-			if((v->languages[0] != 0) && (strcmp(&v->languages[1],"variant") != 0) && (memcmp(v->identifier,"mb/",3) != 0))
+			if((v->languages[0] != 0) && (strcmp(&v->languages[1],"variant") != 0)
+				&& (memcmp(v->identifier,"mb/",3) != 0) && (memcmp(v->identifier,"test/",5) != 0))
 			{
 				voices[j++] = v;
 			}