Merge branch 'development'

dictsource/an_list

@@ -107,57 +107,67 @@ _1M2	un||mil^'on
 _0M4	bil^'ons
 _1M4	un||bil^'on
 
-_0and	i
+_0and	%i
 _dpt	kon
 
 
 // ordinal numbers
-_#º	'eno
-_#ª	'ena
+_#º	o
+_#ª	a
 //_ord	eno
 
-_1o	un
-1o	prim'Er
-_2o	dos
-2o	seQ'und
-_3o	tres
-3o	tErT'Er
-_4o	kwatr
-_5o	Tink
-_6o	seIs
-_7o	sEt
-_8o	gwit
-_9o	n'U
-_10o	DET
-_11o	onT
-_12o	DoT
-_13o	tReT
-_14o	katorT
-_15o	kinT
-_16o	sET
+_1ox	prim'Er
+_2ox	seQ'und
+_3ox	tErT'Er
+
+_1o	un'en
+_2o	dos'en
+_3o	tres'en
+_4o	kwatr'en
+_5o	Tink'en
+_6o	seIs'en
+_7o	sEt'en
+_8o	gwit'en
+_9o	nU'en
+_11o	onT'en
+_12o	DoT'en
+_13o	tReT'en
+_14o	katorT'en
+_15o	kinT'en
+_16o	sET'en
+_10o	DET'en
+_20o	Bint'en
+_30o	trent'en
+_40o	kwarant'en
+_50o	Tinkwant'en
+_60o	siSant'en
+_70o	sEtant'en
+_80o	gwitant'en
+_90o	noBant'en
+
 _1Xo	DeTi
-_20o	Bint
 _2Xo	Binti
-_30o	trent
 _3Xo	trentaI
-_40o	kwarant
 _4Xo	kwarantaI
-_50o	Tinkwant
 _5Xo	TinkwantaI
-_60o	siSant
 _6Xo	siSantaI
-_70o	sEtant
 _7Xo	sEtantaI
-_80o	gwitant
 _8Xo	gwitantaI
-_90o	noBant
 _9Xo	noBantaI
-_0Co	TEnt
-_1Co	TEnt
-_0M1o	mil
-_1M1	mil
-_0M2o	mil^on
-_1M2o	mil^on
+_1Co	TEnt'en
+_2Co	dosTEnt'en
+_3Co	tresTEnt'en
+_4Co	kw,atroTEnt'en
+_5Co	T,inkoTEnt'en
+_6Co	seIsTEnt'en
+_7Co	sjEtTEnt'en
+_8Co	gw,eItoTEnt'en
+_9Co	nweUTEnt'en
+
+_0M1o	mil'en
+_1M1	mil'en
+_0M2o	mil^on'en
+_1M2o	mil^on'en
 
 
 // unstressed words

dictsource/bg_list

@@ -4,13 +4,17 @@
 
 // words of one letter
 в	$combine
-а	$u $combine
+а	$u  $combine
 е	$u
-й i $u
+й	i $u
 и	$u $pause
-у  $u $combine
-с $combine
-г god'ina $hasdot
+у	$u $combine
+с	$combine
+г	god'ina $hasdot
+
+(а в)   %av
+(а с)   %as
+
 
 // letter names
 б	b@

dictsource/en_rules

@@ -5864,7 +5864,7 @@
          _) zs (Y        z
          _) z (wY        z
 
-         n) z (_         ts
+         n) z (_N        ts
          @) zales (_     z'A:l@s
             zei          zaI
             zes (_       zI2z

dictsource/hi_list

@@ -195,12 +195,7 @@ _dpt      _d@s@ml'o:_        // ?? what is Hi for "." ?
 ॐ	#X1o~m
 
 _??	Vks.@r
-_a	ej
-_b	bi:
-_c	si:
-_d	di:
-_e	i:
-_f	Ef
+
 
 // Punctuation
 ।	dVn.d.V

dictsource/kn_list

@@ -26,7 +26,7 @@ p	pi:
 q	kju:
 r	a:r
 s	Es
-t	thi:
+t	t#i:
 u	ju:
 v	vi:
 w	dabalju:

dictsource/ml_list

@@ -63,7 +63,7 @@ p	pi:
 q	kju:
 r	a:R
 s	Es
-t	ti:
+t	t#i:
 u	ju:
 v	vi:
 w	dVbVlju:

dictsource/pt_list

@@ -734,6 +734,7 @@ comodo		$alt2
 concerto	$alt2 $noun
 concordo	$alt
 conforto	$alt  $verb
+congelo		$alt
 consolo		$alt2 $noun
 controle	$alt2 $noun
 controlo	$alt2 $noun
@@ -861,7 +862,6 @@ hemorroida	$alt
 hoje		$alt2
 horrores	$alt2
 ideia		$alt
-insonoro	$alt
 interesse	$alt2 $noun
 jogo		$alt2 $noun
 joguete		$alt2
@@ -993,6 +993,7 @@ selvagem	seUv'aZeIN
 sexteto		$alt2
 ?1 senhora	$alt2
 senhores	$alt2
+singelo		$alt
 sinopse		$alt
 soco		$alt2
 soco		$alt $verb
@@ -1030,6 +1031,7 @@ trombeta	$alt2
 tumores		$alt2
 valeta		$alt2
 vanessa		$alt2
+velo		$alt
 velozes		$alt
 verme		$alt
 violeta		$alt2

dictsource/pt_rules

@@ -359,12 +359,11 @@
  _perd) e (L05_    e
  _quis) e (L05_    E
  _soub) e (L05_    E
-proteg) e (L05_    e         // des-
-  diss) e (L06_    E         / contra-
+  diss) e (L06_    E         // contra-
   tend) e (L06_    e         // a- con- en- es-
+  vend) e (L06_    e         // re-
  _mord) e (L06_    e
  _rend) e (L06_    e
- _vend) e (L06_    e
  prend) e (L06_    e         // a- rea- des- desa-
 _arrepend) e (L06_ e
 _defend) e (L06_   e
@@ -373,7 +372,7 @@ _escond) e (L06_   e
 preend) e (L06_    e         // com- a- re- sur-
 respond) e (L06_   e         // cor-
    bat) e (L07_    e         // re- com-
-   ceb) e (L07_    e         / re- per- con-
+   ceb) e (L07_    e         // re- per- con-
    ced) e (L07_    e         // pro- con- su-
    met) e (L07_    e         // co- pro- arre- re- intro- compro- sub-
    mov) e (L07_    e         // re- co-
@@ -388,6 +387,7 @@ respond) e (L06_   e         // cor-
   tend) e (L07_    e         // en-  con- a- es-
   torc) e (L07_    e         // con-
   venc) e (L07_    e         // con-
+  vend) e (L07_    e         // re-
   vert) e (L07_    e         // in- re- con-
   volv) e (L07_    e         // de- en- re-
  _eleg) e (L07_    e
@@ -401,7 +401,6 @@ respond) e (L06_   e         // cor-
  parec) e (L07_    e         // a- rea- com- desa-
  prend) e (L07_    e         // a- rea- des-
  screv) e (L07_    e         // e- in- ree- pre- de-
-_abastec) e (L07_  e
 _aborrec) e (L07_  e
 _absolv) e (L07_   e
 _adoec) e (L07_    e
@@ -434,6 +433,7 @@ _perec) e (L07_    e
 _rejuvenec) e (L07_ e
 _resolv) e (L07_   e
 _suspend) e (L07_  e
+abastec) e (L07_   e         // re-
 conhec) e (L07_    e         // & re- des-
 estabelec) e (L07_ e         // re-
 favorec) e (L07_   e         // des-
@@ -458,10 +458,12 @@ respond) e (L07_   e         // cor-
   _sev) e (rL03_   E
  _reit) e (rL03_   E
  _sinc) e (rL03_   E
-     g) e (rL04_   E         // su- refri-
+    _g) e (rL04_   E
     qu) e (rL04_   E         // re-
    _op) e (rL04_   E
+   gen) e (rL04_   E         // de- re-
   _alt) e (rL04_   E
+  _sug) e (rL04_   E
   _sup) e (rL04_   E
   _tol) e (rL04_   E
   _ven) e (rL04_   E
@@ -469,6 +471,7 @@ respond) e (L07_   e         // cor-
  _coop) e (rL04_   E
  _temp) e (rL04_   E
 _recup) e (rL04_   E
+_refrig) e (rL04_  E
     _v) e (rbL03_  E
     _p) e (rdL02_  E
     _l) e (rdL03_  E
@@ -503,11 +506,13 @@ _obsol) e (tL03_   e
    scr) e (vL03_   e         // e- in- pre- ree- de-
   _atr) e (vL03_   e
     _r) e (zL04_   E
-    pr) e (zL04_   E         / des- menos-
+   _pr) e (zL04_   E
+_despr) e (zL04_   E
 _embel) e (zL04_   E
+_menospr) e (zL04_ E
 _emagr) e (çL03 
     pr) e (çL03_   e         // a-
-   aqu) e (çL03_   e         / re-
+   aqu) e (çL03_   e         // re-
    mer) e (çL03_   e         // des-
    pad) e (çL03_   e         // com-
    par) e (çL03_   e         // a- re- com- rea- desa-
@@ -859,6 +864,10 @@ _desabr) o (chL04_ O
   _ens) o (pL04_   O
     _s) o (prL04_  O
      _) o (ptL04_  O
+ _inod) o (rL03_   O
+_inson) o (rL03_   O
+_comem) o (rL04    O
+_remem) o (rL04_   O
   _enf) o (rquL02_ O
      c) o (rrL01_  o         // escorra, incorra, corra, recorra
     _m) o (rrL01_  o

dictsource/ta_list

@@ -68,7 +68,7 @@ p	pi:
 q	kju:
 r	a:R
 s	Es
-t	thi:
+t	t#i:
 u	ju:
 v	vi:
 w	dVbVlju:

dictsource/ta_rules

@@ -950,6 +950,7 @@ _துப்பாக்கி) க்க (ுண்ட	gg
     _ப) சக்		tSVk
      _) ச (க்கரவர்த்தி	tSV
      _) ச (க்கிலி	tSV
+     _) சக் (ரவர்த்தி	tSVk
     _க) ச (க்ஸ்தான	zV
         சங்க (ர		s.VNkV
      _) ச (ட்டL08ஸ்கர	tSV
@@ -1057,6 +1058,7 @@ _தணிகா) ச (ல		tSV
  _வரதா) ச (ாரி		tS
  _வேதா) ச (ாரி		tS
 _கிருஷ்ணமா) ச (ாரி	tS
+_கோபால) ச (ாரி		tS
 _சங்கரா) ச (ாரி		tS
 _தேசிகா) ச (ாரி		tS
 _பட்டா) ச (ாரி		tS
@@ -1064,6 +1066,7 @@ _பிரம்ம) ச (ாரி		tS
 _மத்வா) ச (ாரி		tS
 அனந்தா) ச (ாரி		tS
 அன்னமா) ச (ாரி		tS
+ராஜகோபால) ச (ாரி	tS
 ராமானுஜா) ச (ாரி		tS
     ஆ) ச (ாரிய_		tS
     ஆ) ச (ாரியார	tS
@@ -2601,6 +2604,10 @@ _டீயும்_) ப (ன்னும்_	bV
      _) ப (ாகீரதி		b
      _) ப (ாக்கிக்	b
      _) ப (ாக்கிய	b
+     _) ப (ாக்டரி_	f
+     _) ப (ாக்டரியா	b
+     _) ப (ாக்டரியி	f
+     _) ப (ாக்டரியை	f
      _) பாக்ட (ீரியா	ba:kt.
      _) பாக்த (ாத		ba:gd
      _) ப (ாக்ய		b
@@ -2947,6 +2954,7 @@ _விளையாடு_) ப (ாம்பே	p
      _) பி (ராதித்	p
      _) பி (ராத்தல	b
      _) பி (ராந்தி_	b
+     _) பி (ராந்திய_	p
      _) பி (ராந்தியங்	p
      _) பி (ராந்தியத்	p
      _) பி (ராந்தியம்_	p
@@ -2986,8 +2994,10 @@ _விளையாடு_) ப (ாம்பே	p
      _) பி (ரித்தானி	b
      _) பிரின்சிப (ால	prinsip
      _) பி (ரிமிய	p
+     _) பி (ரியங்கள	p
      _) பிரியங்க (ா	prijVNk
      _) பி (ரியதர	p
+     _) பி (ரியத்த	p
      _) பி (ரியன	p
      _) பி (ரியப்பட	p
      _) பி (ரியம	p

dictsource/te_list

@@ -24,7 +24,7 @@ p	pi:
 q	kju:
 r	a:r
 s	Es
-t	thi:
+t	t#i:
 u	ju:
 v	vi:
 w	dVbVlju:

dictsource/vi_rules

@@ -77,7 +77,7 @@ ỵ	ỵ
 //================================
 //  Consonants
 //================================
-.L01 t c ne nd k x s nt ch th me te le ce ny ry e o u re nch de p pe ph sh v ve 
+.L01 t c ne nd k x s nt ch th me te le ce ny ry e o u re nch de p pe ph sh v ve ge
 .L02 n
 
 
@@ -460,7 +460,7 @@ ỵ	ỵ
 
 .group i
         i          i
-        i (A       j
+        //i (A       j
         ia         i@
         ià         i@2
         iá         i@3
@@ -822,7 +822,7 @@ ỵ	ỵ
 
 .group y
         y          i
-        y (A       j
+       // y (A       j
         ya         i@
         yà         i@2
         yá         i@3
@@ -837,7 +837,7 @@ ỵ	ỵ
         yệ         i@6
       //  yêu        i@U
         
- 
+        y (L01     _^_EN
 
 .group ỳ
         ỳ          i2

src/compiledata.cpp

@@ -745,6 +745,7 @@ static void CompileReport(void)
 	REF_HASH_TAB **list;
 	const char *data_path;
 	int prev_table;
+	int procedure_num;
 	int prev_mnemonic;
 
     if(f_report == NULL)
@@ -802,7 +803,17 @@ static void CompileReport(void)
 			j++;
 		}
 
-		fprintf(f_report,"  [%s] %s",WordToString(prev_mnemonic = list[ix]->ph_mnemonic), phoneme_tab_list2[prev_table = list[ix]->ph_table].name);
+		prev_mnemonic = list[ix]->ph_mnemonic;
+		if((prev_mnemonic >> 24) == 'P')
+		{
+			// a procedure, not a phoneme
+			procedure_num = atoi(WordToString(prev_mnemonic));
+			fprintf(f_report,"  %s  %s", phoneme_tab_list2[prev_table = list[ix]->ph_table].name, proc_names[procedure_num]);
+		}
+		else
+		{
+			fprintf(f_report,"  [%s] %s",WordToString(prev_mnemonic), phoneme_tab_list2[prev_table = list[ix]->ph_table].name);
+		}
 		fputc('\n',f_report);
 	}
 
@@ -1856,7 +1867,7 @@ static int LoadWavefile(FILE *f, const char *fname)
         }
 
         sprintf(command,"sox \"%s%s.wav\" -r %d -c1 -t wav %s\n",path_source,fname2,samplerate_native, fname_temp);
-        if(system(command) < 0)
+        if(system(command) != 0)
         {
             failed = 1;
         }
@@ -2761,6 +2772,8 @@ int CompilePhoneme(int compile_phoneme)
 		}
 		strcpy(proc_names[n_procs], item_string);
 		phoneme_out = &phoneme_out2;
+		sprintf(number_buf,"%.3dP", n_procs);
+		phoneme_out->mnemonic = StringToWord(number_buf);
 	}
 
 	phoneme_out->code = phcode;
@@ -3498,10 +3511,11 @@ static void CompilePhonemeData2(const char *source)
 make_envs();
 #endif
 
+	wxLogStatus(_T("Compiling phoneme data: ")+wxString(path_source,wxConvLocal));
 	n_envelopes = 0;
 	error_count = 0;
 	resample_count = 0;
-memset(markers_used,0,sizeof(markers_used));
+	memset(markers_used,0,sizeof(markers_used));
 
 	f_errors = stderr;
 
@@ -3668,7 +3682,7 @@ fprintf(f_errors,"\nRefs %d,  Reused %d\n",count_references,duplicate_references
 
 	if(error_count > 0)
 	{
-		report += _T(" See file: 'phsource/error_log'.");
+		report += _T(" See file: '")+wxString(path_source,wxConvLocal)+_T("phsource/error_log'.");
 		wxLogError(report);
 	}
 	wxLogStatus(report + report_dict);
@@ -4138,7 +4152,11 @@ void CompilePhonemeData()
 void CompileSampleRate()
 {
     long value;
-    value = wxGetNumberFromUser(_T("Compile phoneme data with a specified sample rate"), _T("Sample rate"), _T("Resample (needs sox)"), 22050, 5000, 48000);
+#ifndef PLATFORM_POSIX
+	wxLogError(_T("Change Sample Rate needs the 'sox' program.  It probably doesn't work on Windows"));
+#endif
+
+    value = wxGetNumberFromUser(_T("Compile phoneme data with a specified sample rate"), _T("Sample rate"), _T("Resample (needs 'sox' program)"), 22050, 5000, 48000);
 
     if(value > 1000)
     {

src/menus.cpp

@@ -91,8 +91,8 @@ wxMenuBar *MakeMenu(int type, const char *dict_name)
         data_menu = new wxMenu;
         data_menu->Append(MENU_COMPILE_DICT, _("Compile &dictionary"));
         data_menu->Append(MENU_COMPILE_DICT_DEBUG, _("Compile dictionary (debu&g)"));
-        data_menu->Append(MENU_COMPILE_PH, _("Compile &phoneme data"));
-        data_menu->Append(MENU_COMPILE_PH2, _("Change sample &rate"));
+        data_menu->Append(MENU_COMPILE_PH, _("Compile &phoneme data 22050Hz"));
+        data_menu->Append(MENU_COMPILE_PH2, _("Compile at sample &rate"));
         data_menu->AppendSeparator();
         data_menu->Append(MENU_COMPILE_MBROLA, _("Compile &mbrola phonemes list..."));
         data_menu->Append(MENU_COMPILE_INTONATION, _("Compile &intonation data"));

src/synthdata.cpp

@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.46.51  15.Mar.13";
+const char *version_string = "1.46.53  16.Mar.13";
 const int version_phdata  = 0x014640;
 
 int option_device_number = -1;

src/synthesize.cpp

@@ -82,7 +82,7 @@ const char *WordToString(unsigned int word)
 	int  ix;
 	static char buf[5];
 
-	for(ix=0; ix<3; ix++)
+	for(ix=0; ix<4; ix++)
 		buf[ix] = word >> (ix*8);
 	buf[4] = 0;
 	return(buf);

src/tr_languages.cpp

@@ -455,6 +455,7 @@ Translator *SelectTranslator(const char *name)
 	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 short stress_lengths_ta2[8] = {230, 230,  240, 240,  0, 0,  260, 260};
 	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()
@@ -731,6 +732,7 @@ 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;
 			}
 			else
 			if(name2 == L_pap)
@@ -1423,7 +1425,7 @@ SetLengthMods(tr,3);  // all equal
 	case L('m','r'):  // Marathi
 	case L('t','e'):  // Telugu
 		{
-			SetupTranslator(tr,stress_lengths_ta,stress_amps_ta);
+			SetupTranslator(tr,stress_lengths_ta2, 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;
@@ -1433,9 +1435,12 @@ SetLengthMods(tr,3);  // all equal
 
 			if(name2 == L('t','a'))
 			{
+				SetupTranslator(tr,stress_lengths_ta, NULL);
 				tr->letter_bits_offset = OFFSET_TAMIL;
 				tr->langopts.numbers =  NUM_OMIT_1_THOUSAND ;
+				tr->langopts.param[LOPT_WORD_MERGE] = 1;   // don't break vowels betwen words
 			}
+			else
 			if(name2 == L('m','r'))
 			{
 				tr->letter_bits_offset = OFFSET_DEVANAGARI;
@@ -1458,9 +1463,8 @@ SetLengthMods(tr,3);  // all equal
 				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_
-				SetLetterBitsRange(tr,LETTERGP_B,0x4e,0x4e);   // chillu-virama (unofficial)
+			SetLetterBitsRange(tr,LETTERGP_B,0x4e,0x4e);   // chillu-virama (unofficial)
 		}
 		break;
 

src/translate.cpp

@@ -842,7 +842,7 @@ int TranslateWord(Translator *tr, char *word_start, int next_pause, WORD_TAB *wt
 	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
+	int c_temp;   // save a character byte while we temporarily replace it with space
 	int first_char;
 	int last_char = 0;
 	int add_plural_suffix = 0;
@@ -912,7 +912,9 @@ int TranslateWord(Translator *tr, char *word_start, int next_pause, WORD_TAB *wt
 	if((word_length == 1) && (wflags & FLAG_TRANSLATOR2))
 	{
 		// retranslating a 1-character word using a different language, say its name
-		spell_word = 1;
+		utf8_in(&c_temp, wordx+1);  // the next character
+		if(!IsAlpha(c_temp) || (AlphabetFromChar(last_char) != AlphabetFromChar(c_temp)))
+			spell_word = 1;
 	}
 
 	if(option_sayas == SAYAS_KEY)
@@ -1938,6 +1940,7 @@ static int TranslateWord2(Translator *tr, char *word, WORD_TAB *wtab, int pre_pa
 			if(ok)
 			{
 				*p2 = '-'; // replace next space by hyphen
+				wtab[0].flags &= ~FLAG_ALL_UPPER;  // prevent it being considered an abbreviation
 				flags = TranslateWord(translator, word, next_pause, wtab, NULL);  // translate the combined word
 				if((sylimit > 0) && (CountSyllables(p) > (sylimit & 0x1f)))
 				{