[1.31.09]

More fixes for &--; sequences.
lang=es: use phoneme [O] for letter 'o' except at end of word.
leng=es: attempt to improve single 'r' between two vowels.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@139 d46cf337-b52f-0410-862d-fd96e6ae7743

src/intonation.cpp

@@ -275,12 +275,11 @@ static TONE_HEAD tone_head_table[N_TONE_HEAD_TABLE] = {
 };
 
 static TONE_NUCLEUS tone_nucleus_table[N_TONE_NUCLEUS_TABLE] = {
-   {PITCHfall,   30, 5,  PITCHfall,   30, 8, NULL, 12, 7, 0},      // 0 statement
+   {PITCHfall,   30, 5,  PITCHfall,   32, 9, NULL, 12, 7, 0},      // 0 statement
    {PITCHfrise,  35, 8,  PITCHfrise2, 35,10, NULL, 15, 23, 0},     // 1 comma
    {PITCHfrise,  39,10,  PITCHfrise2, 36,10, NULL, 15, 28, 0},     // 2 question
 //   {PITCHfall,   41, 4,  PITCHfall,   41,27, NULL, 16, 4, T_EMPH},  // 3 exclamation
-   {PITCHfall,   43, 3,  PITCHfall,   43,35, NULL, 35, 4, T_EMPH},  // 3 exclamation
-//   {PITCHrisefall,47, 8, PITCHrise,   48,30, NULL, 48, 5, T_EMPH},  // 3 exclamation
+   {PITCHfall,   41, 4,  PITCHfall,   41,35, NULL, 35, 4, T_EMPH},  // 3 exclamation
    {PITCHfall,   38, 2,  PITCHfall,   42,30, NULL, 15, 5, 0},      // 4 statement, emphatic
    {PITCHfall,   28, 5,  PITCHfall,   28, 9, NULL, 12, 7, 0},      // 5 statement, less intonation
    {PITCHfrise,  30, 8,  PITCHfrise2, 30,10, NULL, 13, 20, 0},      // 6 comma, less intonation
@@ -772,6 +771,7 @@ void Translator::CalcPitches_Tone(int clause_tone)
 		if((p->type == phPAUSE) && (p->ph->std_length > 50))
 		{
 			pause = 1;  // there is a pause since the previous vowel
+			prevw_tph = phoneme_tab[phonPAUSE];  // forget previous tone
 		}
 
 		if(p->newword)
@@ -817,7 +817,7 @@ void Translator::CalcPitches_Tone(int clause_tone)
 					}
 				}
 
-				if(prev_tph->mnemonic == 0x343132)  // [214]
+				if(prevw_tph->mnemonic == 0x343132)  // [214]
 				{
 					if(tph->mnemonic == 0x343132)   // [214]
 						prev_p->tone_ph = LookupPh("35");

src/readclause.cpp

@@ -1695,6 +1695,7 @@ MNEM_TAB xml_char_mnemonics[] = {
 	{"amp", '&'},
 	{"quot", '"'},
 	{"nbsp", ' '},
+	{"apos", '\''},
 	{NULL,-1}};
 
 
@@ -1723,13 +1724,18 @@ int Translator::ReadClause(FILE *f_in, char *buf, unsigned short *charix, int n_
 	int n_xml_buf;
 	int terminator;
 	int punct;
+	int found;
 	int any_alnum = 0;
 	int self_closing;
 	int punct_data;
 	const char *p;
-	char buf2[40];
 	wchar_t xml_buf[N_XML_BUF+1];
 
+#define N_XML_BUF2   12
+	char buf2[N_XML_BUF2+2];
+	static char ungot_string[N_XML_BUF2+4];
+	static int ungot_string_ix = -1;
+
 	if(clear_skipping_text)
 	{
 		skipping_text = 0;
@@ -1752,14 +1758,13 @@ f_input = f_in;  // for GetC etc
 	if(ungot_char2 != 0)
 	{
 		c2 = ungot_char2;
-		ungot_char2 = 0;
 	}
 	else
 	{
 		c2 = GetC();
 	}
 
-	while(!Eof() || (ungot_char != 0))
+	while(!Eof() || (ungot_char != 0) || (ungot_char2 != 0) || (ungot_string_ix >= 0))
 	{
 		if(!iswalnum(c1))
 		{
@@ -1782,47 +1787,79 @@ f_input = f_in;  // for GetC etc
 
 		cprev = c1;
 		c1 = c2;
-		c2 = GetC();
 
-		if(Eof())
+		if(ungot_string_ix >= 0)
 		{
-			c2 = ' ';
+			if(ungot_string[ungot_string_ix] == 0)
+				ungot_string_ix = -1;
 		}
 
+		if((ungot_string_ix == 0) && (ungot_char2 == 0))
+		{
+			c1 = ungot_string[ungot_string_ix++];
+		}
+		if(ungot_string_ix >= 0)
+		{
+			c2 = ungot_string[ungot_string_ix++];
+		}
+		else
+		{
+			c2 = GetC();
+
+			if(Eof())
+			{
+				c2 = ' ';
+			}
+		}
+		ungot_char2 = 0;
+
 		if((option_ssml) && (phoneme_mode==0))
 		{
 			if((ssml_ignore_l_angle != '&') && (c1 == '&') && ((c2=='#') || ((c2 >= 'a') && (c2 <= 'z'))))
 			{
 				n_xml_buf = 0;
 				c1 = c2;
-				while(!Eof() && (iswalnum(c1) || (c1=='#')) && (n_xml_buf < 12))
+				while(!Eof() && (iswalnum(c1) || (c1=='#')) && (n_xml_buf < N_XML_BUF2))
 				{
 					buf2[n_xml_buf++] = c1;
 					c1 = GetC();
 				}
 				buf2[n_xml_buf] = 0;
 				c2 = GetC();
+				sprintf(ungot_string,"%s%c%c",&buf2[0],c1,c2);
 
-				if(buf2[0] == '#')
+				if(c1 == ';')
 				{
-					// character code number
-					c1 = '#';  // in case there isn't a number
-					if(buf2[1] == 'x')
-						sscanf(&buf2[2],"%x",(unsigned int *)(&c1));
+					if(buf2[0] == '#')
+					{
+						// character code number
+						if(buf2[1] == 'x')
+							found = sscanf(&buf2[2],"%x",(unsigned int *)(&c1));
+						else
+							found = sscanf(&buf2[1],"%d",&c1);
+					}
 					else
-						sscanf(&buf2[1],"%d",&c1);
+					{
+						if((found = LookupMnem(xml_char_mnemonics,buf2)) != -1)
+						{
+							c1 = found;
+							if(c2 == 0)
+								c2 = ' ';
+						}
+					}
 				}
 				else
 				{
-					if((j = LookupMnem(xml_char_mnemonics,buf2)) != -1)
-					{
-						c1 = j;
-					}
-					else
-					{
-						c1 = '&';
-					}
+					found = -1;
 				}
+
+				if(found <= 0)
+				{
+					ungot_string_ix = 0;
+					c1 = '&';
+					c2 = ' ';
+				}
+
 				if((sayas_mode == 0x14) && (c1 <= 0x20))
 				{
 					c1 += 0xe000;  // move into unicode private usage area

src/setlengths.cpp

@@ -310,6 +310,9 @@ void Translator::CalcLengths()
 			else
 				p->prepause = 60;
 
+			if(prev->type == phSTOP)
+				p->prepause = 60;
+
 			if((langopts.word_gap & 0x10) && (p->newword))
 				p->prepause = 60;
 

src/synthdata.cpp

@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.31.08  08.Feb.08";
+const char *version_string = "1.31.09  11.Feb.08";
 const int version_phdata  = 0x013105;
 
 int option_device_number = -1;