From fb7230a9122b07a42b68e12d8a1abfc2d506e331 Mon Sep 17 00:00:00 2001
From: jonsd <jonsd@d46cf337-b52f-0410-862d-fd96e6ae7743>
Date: Thu, 20 Dec 2007 13:45:16 +0000
Subject: [PATCH] [1.29.29] Add inital draft of language "ta", Tamil lang=en:
 Don't treat "Mc" prefix as a separate word, combine it with the following
 word.

git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@113 d46cf337-b52f-0410-862d-fd96e6ae7743
---
 dictsource/dict_phonemes |   4 +
 dictsource/en_list       |   2 +-
 dictsource/en_rules      |   5 +-
 dictsource/mk_list       |  32 +--
 dictsource/ta_list       |   4 +
 dictsource/ta_rules      | 173 +++++++++++++++++
 espeak-data/voices/grc   |   2 +-
 espeak-data/voices/ta    |   4 +
 phsource/compile_report  |  26 ++-
 phsource/ph_serbian      |  43 ++++-
 phsource/ph_tamil        |  84 ++++++++
 phsource/ph_zh           | 407 +++++++++++++++++++++++++++++++++++++++
 phsource/phonemes        |   3 +
 phsource/ufric/sh_sr.wav | Bin 4652 -> 5828 bytes
 phsource/vowel/a#_2      | Bin 6240 -> 7276 bytes
 src/dictionary.cpp       |  31 +--
 src/intonation.cpp       |  15 +-
 src/readclause.cpp       |  11 +-
 src/synthdata.cpp        |   2 +-
 src/synthesize.cpp       |   3 -
 src/synthesize.h         |   2 +-
 src/tr_english.cpp       |   3 +-
 src/tr_languages.cpp     |  34 +++-
 src/translate.cpp        |  19 +-
 src/translate.h          |   2 +-
 25 files changed, 819 insertions(+), 92 deletions(-)
 create mode 100644 dictsource/ta_list
 create mode 100644 dictsource/ta_rules
 create mode 100755 espeak-data/voices/ta
 create mode 100644 phsource/ph_tamil
 create mode 100644 phsource/ph_zh

diff --git a/dictsource/dict_phonemes b/dictsource/dict_phonemes
index 5e772566..c17c4317 100644
--- a/dictsource/dict_phonemes
+++ b/dictsource/dict_phonemes
@@ -262,6 +262,10 @@ l    m    n    N    p    r    s    t
 tS   v    w    z    
 
 
+Dictionary ta_dict
+
+
+
 Dictionary pt_dict
 
 &    &/   &U~  &~   @    @-   a    A    
diff --git a/dictsource/en_list b/dictsource/en_list
index d671035d..ed81746e 100644
--- a/dictsource/en_list
+++ b/dictsource/en_list
@@ -169,7 +169,7 @@ i.e	aIi:_!   $pause
 irc	$abbrev
 lbs	paUndz
 ltd	lImItId
-mc	m@k
+mc	m@k  $alt2   // combine with the following word
 oem	$abbrev
 ok	$abbrev
 omg	$abbrev
diff --git a/dictsource/en_rules b/dictsource/en_rules
index d5bae534..91398b00 100644
--- a/dictsource/en_rules
+++ b/dictsource/en_rules
@@ -3227,7 +3227,9 @@
         li) mb (e        m
          o) mb (e        m
         du) mb (e        m
-         _) m (c         m@
+         _) mc           m@k
+         _) mc (A        mak
+         _) mc (-A       mak
             mn (_        m
          _) mn           n
         &B) me (_        mI2
@@ -4965,6 +4967,7 @@
          t) y (rann      I
         ph) ys (i        Iz
             yr (_        @
+            yre (_       aI3
          c) y (ni        I
          l) y (ri        I
          p) y (ra        I
diff --git a/dictsource/mk_list b/dictsource/mk_list
index e35e3524..705babbe 100755
--- a/dictsource/mk_list
+++ b/dictsource/mk_list
@@ -162,37 +162,7 @@ _1M4  bili'on
 _dpt  _:z'api*k&_
 _0and i
 
-// Roman numerals
-ii	$abbrev
-iii	$abbrev
-iv	$abbrev
-vii	$abbrev
-viii	$abbrev
-ix	$abbrev
-xi	$abbrev
-xii	$abbrev
-xiii	$abbrev
-xiv	$abbrev
-xvi	$abbrev
-xvii	$abbrev
-xviii	$abbrev
-xix	$abbrev
-xxi	$abbrev
-xxii	$abbrev
-xxiii	$abbrev
-xxiv	$abbrev
-xxvi	$abbrev
-xxvii	$abbrev
-xxviii	$abbrev
-xxix	$abbrev
-xxxi	$abbrev
-xxxii	$abbrev
-xxxiii	$abbrev
-xxxiv	$abbrev
-xxxvi	$abbrev
-xxxvii	$abbrev
-xxxviii  $abbrev
-xxxix	$abbrev
+
 
 // Abbreviations
 
diff --git a/dictsource/ta_list b/dictsource/ta_list
new file mode 100644
index 00000000..68c79c98
--- /dev/null
+++ b/dictsource/ta_list
@@ -0,0 +1,4 @@
+ 
+// This file is UTF8 encoded 
+// Spelling to phoneme words and exceptions for Tamil
+
diff --git a/dictsource/ta_rules b/dictsource/ta_rules
new file mode 100644
index 00000000..f5518f52
--- /dev/null
+++ b/dictsource/ta_rules
@@ -0,0 +1,173 @@
+
+// This file is UTF8 encoded 
+// Spelling to phoneme rules for Tamil
+
+.replace
+௦	0	// Tamil numbers
+௧	1
+௨	2
+௩	3
+௪	4
+௫	5
+௬	6
+௭	7
+௮	8
+௯	9
+
+
+.group 0xe0ae
+	ஂ		// anusvara
+
+	ஃ		// visarga
+
+	அ	a
+
+	ஆ	a:
+
+	இ	i
+
+	ஈ	i:
+
+	உ	u
+
+	ஊ	u:
+
+	எ	e
+
+	ஏ	e:
+
+	ஐ	E:
+
+	ஒ	o
+
+	ஓ	o:
+
+	ஔ	aU
+
+// consonants
+
+	க	ga   // inter-vocalic, unless there is virama before or after
+	க (B	g
+	க (் 	g
+     ்) க	ga
+     ்) க (B	g
+     _) க	ka
+     _) க (B	k
+	க்க	ka
+	க்க (B	k
+
+	ங	Na
+	ங (B	N
+
+	ச	dZa
+	ச (B	dZ
+     _) ச	sa
+     _) ச (B	s
+	ச்ச	tSa
+	ச்ச (B	tS
+
+	ஜ	dZa
+	ஜ (B	dZ
+
+	ஞ	n^a
+	ஞ (B	n^
+
+	ட	d.a
+	ட (B	d.
+     _) ட	t.a
+     _) ட (B	t.
+	ட்ட	t.a
+	ட்ட (B	t.
+
+	ண	n.a
+	ண (B	n.
+
+	த	da
+	த (B	d
+     _) த	ta
+     _) த (B	t
+	த்த	ta
+	த்த (B	t
+
+	ந	na
+	ந (B	n
+
+	ன	na
+	ன (B	n
+
+	ப	ba
+	ப (B	b
+     _) ப	pa
+     _) ப (B	p
+	ப்ப	pa
+	ப்ப (B	p
+
+	ம	ma
+	ம (B	m
+
+	ய	ja
+	ய (B	j
+
+	ர	ra
+	ர (B	r
+
+	ற	Ra
+	ற (B	R
+	ற் (ற	t	// RR -> tR
+
+	ல	la
+	ல (B	l
+
+	ள	l/2a    // this should be [l.] not [l/2]
+	ள (B	l/2
+
+	ழ	z.a
+	ழ (B	z.
+
+	வ	va
+	வ (B	v
+
+	ஶ	Sa
+	ஶ (B	S
+
+	ஷ	s.a
+	ஷ (B	s.
+
+	ஸ	sa
+	ஸ (B	s
+
+	ஹ	ha
+	ஹ (B	h
+
+
+// combining vowels
+
+	ா	a:
+
+	ி	i
+
+.group 0xe0af
+	ீ	i:
+
+	ு	u
+
+	ூ	u:
+
+	ெ	e
+
+	ே	e:
+
+	ை	E:
+
+	ொ	o
+
+	ோ	o:
+
+	ௌ	aU
+
+	்		// virama
+
+	ௗ	:	// aU length mark
+
+.group
+	$	dola
diff --git a/espeak-data/voices/grc b/espeak-data/voices/grc
index a8799222..ffa94206 100755
--- a/espeak-data/voices/grc
+++ b/espeak-data/voices/grc
@@ -4,5 +4,5 @@ gender male
 
 stressLength 170 170  190 190  0 0  230 240
 dictrules 1
-words 1
+words 3
 
diff --git a/espeak-data/voices/ta b/espeak-data/voices/ta
new file mode 100755
index 00000000..43dd4af9
--- /dev/null
+++ b/espeak-data/voices/ta
@@ -0,0 +1,4 @@
+name tamil-test
+language ta
+gender male
+
diff --git a/phsource/compile_report b/phsource/compile_report
index 0f680da2..af7abf95 100644
--- a/phsource/compile_report
+++ b/phsource/compile_report
@@ -1,4 +1,4 @@
-41 phoneme tables
+42 phoneme tables
           new total
     base  96   96
    base2  24  115
@@ -16,6 +16,7 @@
       fr  34  118
    fr_ca  11  118
       hi  50  132
+      ta  12  134
       hu  23  112
       nl  26  117
       pl  15  107
@@ -23,7 +24,7 @@
       cs   6  125
       hr  20  130
       mk   3  131
-      sr  11  130
+      sr  14  130
       ru  38  126
       it  17  118
       la  21  114
@@ -427,6 +428,7 @@ r3/r_trill      [R2] base
                 [r] cy
                 [r] nl
                 [R] sk
+                [r*] sr
                 [x] pt
 r3/r_trill2     [R] base
                 [R] hr
@@ -438,6 +440,7 @@ r3/r_trill3.wav  [R3] base
                 [r] nl
 r3/r_trill.wav  [R2] base
                 [R] sk
+                [r*] sr
                 [x] pt
 r3/r_u          [(u)] base
 r3/rx           [*] base
@@ -699,6 +702,7 @@ ustop/t_dnt     [t] base
                 [t] hu
                 [t] pl
                 [t] hr
+                [t] sr
                 [t] ru
                 [t;] ro
                 [t[] vi
@@ -741,7 +745,6 @@ ustop/ts_pzd_   [c] hi
 ustop/ts_pzd2   [c] hi
                 [c] hu
 ustop/t_sr      [d] sr
-                [t] sr
 ustop/ts_rfx    [ts.] zh
 ustop/ts_rfx_unasp  [ts.] zh
 ustop/ts_sr.wav  [tS;] sr
@@ -1106,6 +1109,7 @@ vowel/a         [a] en_n
                 [a] zh
 vowel/a#        [a/] base2
                 [a2] en_sc
+                [&] sr
                 [&] pt
                 [&/] pt
                 [a2] en_wi
@@ -1120,7 +1124,8 @@ vowel/a_2       [a] base2
                 [A:] en_wi
                 [A@] en_wi
                 [aa] en_wi
-vowel/a#_2      [&] hr
+vowel/a#_2      [a] ta
+                [&] hr
                 [a2] sv
                 [&] is
                 [a/] sw
@@ -1150,6 +1155,7 @@ vowel/a_5       [a/] it
                 [a] sw
 vowel/aa        [a] fi
                 [A] fr_ca
+                [a:] ta
                 [A] no
                 [A:] no
                 [aa] zhy
@@ -1185,6 +1191,8 @@ vowel/e         [e] base2
                 [e] eo
                 [e] fr
                 [E:] fr
+                [e] ta
+                [e:] ta
                 [e] hr
                 [e/] it
                 [e:] la
@@ -1267,6 +1275,7 @@ vowel/i         [i] base2
                 [i:] cy
                 [i] eo
                 [I] fr
+                [i:] ta
                 [i:] hu
                 [i] pl
                 [i] it
@@ -1301,6 +1310,7 @@ vowel/i#_5      [y] pt_pt
                 [i/] pt_pt
                 [y] ro
 vowel/i_6       [i:] en_us
+                [i] ta
                 [i] hr
 vowel/i_en      [i:] en
                 [i] vi
@@ -1338,6 +1348,7 @@ vowel/ii_5      [I] en_wi
                 [I2] en_wi
 vowel/ii_6      [I] en_wm
                 [I2] en_wm
+                [I] sr
 vowel/ii_en     [i] en
 vowel/@_low     [3] en_rp
                 [@] en_rp
@@ -1353,6 +1364,8 @@ vowel/o         [o] base2
                 [o:] en
                 [o:] de
                 [o] hi
+                [o] ta
+                [o:] ta
                 [o] it
                 [o/] it
                 [o:] la
@@ -1450,7 +1463,8 @@ vowel/u_2       [u1] fi
                 [u] sk
                 [U] hr
                 [u] ro
-vowel/u#_2      [u-] sv
+vowel/u#_2      [u-] ta
+                [u-] sv
 vowel/u_3       [yU] ro
 vowel/u_4       [u:] en_n
 vowel/u#_4      [U] en_sc
@@ -1465,6 +1479,7 @@ vowel/u_bck     [u] base2
                 [u] fi
                 [u] fr_ca
                 [u:] hi
+                [u:] ta
                 [u:] hu
                 [u] nl
                 [u] pl
@@ -1489,6 +1504,7 @@ vowel/uu_4      [U] fi
 vowel/uu_bck    [U] en_n
                 [U] fr_ca
                 [U] hi
+                [u] ta
                 [u] hu
                 [U] pt
                 [U] no
diff --git a/phsource/ph_serbian b/phsource/ph_serbian
index 26ae637e..d1d5eea4 100644
--- a/phsource/ph_serbian
+++ b/phsource/ph_serbian
@@ -13,14 +13,39 @@
 // Only phonemes which differ from the inherited phonemes need
 // to be included here.
 
+phoneme I
+  vowel starttype (i) endtype (i)
+  length 120
+  formants vowel/ii_6
+  linkout ;
+endphoneme
+
+
+phoneme &
+  vowel starttype (a) endtype (a)
+  length 140
+  formants vowel/a#
+endphoneme
+
+
+phoneme r*   // this is [R] from Slovak/Czech
+  liquid
+  vowelin  f1=0  f2=1700 -300 300  f3=-300 80
+  vowelout f1=2  f2=1700 -300 300  f3=-300 80 brk
+  formants r3/r_trill+r3/r_trill.wav%50
+  trill
+  lengthmod 6
+endphoneme
+
+
 
 phoneme t        // dental variant of /t/
   vls dnt stop
   vowelin f1=0  f2=1600 -300 300  f3=-100 80
   vowelout f1=0 f2=1600 -300 250  f3=-100 80  rms=20
   lengthmod 2
-  wave  ustop/t_sr%50
-  before _ ustop/t_sr%35
+  wave  ustop/t_dnt%50
+  before _ ustop/t_dnt%50
   switchvoicing d
 endphoneme
 
@@ -97,8 +122,8 @@ phoneme tS
   vls pla afr sibilant
   vowelin f1=0  f2=2300 200 400  f3=-100 80
   lengthmod 2
-  wave ustop/tsh_sr.wav
-  before _ ustop/tsh_sr.wav
+  wave ustop/tsh_sr.wav%50
+  before _ ustop/tsh_sr.wav%50
   switchvoicing dZ
 endphoneme
 
@@ -113,12 +138,14 @@ phoneme tS;
 endphoneme
 
 
+
+
 phoneme Z
   vcd pla frc sibilant
   vowelin  f1=0  f2=2000 0 300  f3=-200 80
   vowelout f1=2  f2=2000 0 300  f3=-200 80 brk
-  formants voc/zh+ufric/sh_sr.wav
-  before _ voc/zh+ufric/sh_sr.wav
+  formants voc/zh+ufric/sh_sr.wav%60
+  before _ voc/zh+ufric/sh_sr.wav%50
   lengthmod 6
   switchvoicing S
 endphoneme
@@ -130,8 +157,8 @@ phoneme S
   vowelout f1=0  f2=2100 100 300  f3=-100 80
   length 100
   lengthmod 3
-  wave ufric/sh_sr.wav
-  before _ ufric/sh_sr.wav
+  wave ufric/sh_sr.wav%70
+  before _ ufric/sh_sr.wav%60
   switchvoicing Z
 endphoneme
 
diff --git a/phsource/ph_tamil b/phsource/ph_tamil
new file mode 100644
index 00000000..d6868723
--- /dev/null
+++ b/phsource/ph_tamil
@@ -0,0 +1,84 @@
+
+// Phonemes for Tamil
+// This inherits from ph_hindi
+
+
+
+phoneme a
+  vowel  starttype (@)  endtype (@)
+  length 130
+  formants vowel/a#_2
+endphoneme
+
+
+phoneme a:
+  vowel  starttype (a)  endtype (a)
+  length 270
+  formants vowel/aa
+endphoneme
+
+
+phoneme e
+  vowel  starttype (e)  endtype (e)
+  length 130
+  formants vowel/e
+endphoneme
+
+
+phoneme e:
+  vowel  starttype (e)  endtype (e)
+  length 270
+  formants vowel/e
+endphoneme
+
+
+phoneme i
+  vowel  starttype (i)  endtype (i)
+  length 130
+  formants vowel/i_6
+endphoneme
+
+
+phoneme i:
+  vowel  starttype (i)  endtype (i)
+  length 270
+  formants vowel/i
+endphoneme
+
+
+
+phoneme o
+  vowel  starttype (o)  endtype (o)
+  length 130
+  formants vowel/o
+endphoneme
+
+
+phoneme o:
+  vowel  starttype (o)  endtype (o)
+  length 270
+  formants vowel/o
+endphoneme
+
+
+phoneme u
+  vowel  starttype (u)  endtype (u)
+  length 130
+  formants vowel/uu_bck
+endphoneme
+
+
+phoneme u:
+  vowel  starttype (u)  endtype (u)
+  length 270
+  formants vowel/u_bck
+endphoneme
+
+
+phoneme u-
+  vowel starttype (u) endtype (u)
+  length 130
+  formants vowel/u#_2
+endphoneme
+
+
diff --git a/phsource/ph_zh b/phsource/ph_zh
new file mode 100644
index 00000000..3486d574
--- /dev/null
+++ b/phsource/ph_zh
@@ -0,0 +1,407 @@
+//based on "phonemes"
+//====================================================
+//  Tone Numbers
+//====================================================
+
+phoneme 11   //  tone: low level
+  stress
+  tone 15 10 envelope/p_fall NULL
+endphoneme
+
+phoneme 21   //  tone: low fall
+  stress
+  tone 20 10 envelope/p_fall NULL
+endphoneme
+
+phoneme 214   //  tone: fall rise
+  stress
+  tone 20 40 envelope/p_214 NULL
+endphoneme
+
+phoneme 22   //  tone: mid-low level
+  stress
+  tone 21 19 envelope/p_fall NULL
+endphoneme
+
+phoneme 33   //  tone: mid level
+  stress
+  tone 31 29 envelope/p_fall NULL
+endphoneme
+
+phoneme 35   //  tone: mid rise
+  stress
+  tone 30 50 envelope/p_rise NULL
+endphoneme
+
+phoneme 44   //  tone: mid-high level
+  stress
+  tone 41 39 envelope/p_level NULL
+endphoneme
+
+phoneme 51   //  tone: high fall
+  stress
+  tone 50 10 envelope/p_fall NULL
+endphoneme
+
+phoneme 53   //  tone: high fall
+  stress
+  tone 50 30 envelope/p_fall NULL
+endphoneme
+
+phoneme 55   //  tone: high level
+  stress
+  tone 50 50 envelope/p_level NULL
+endphoneme
+
+//====================================================
+//  Consonants
+//====================================================
+
+phoneme p
+  vls blb stop
+  vowelin  f1=0  f2=1000 -50 -100  f3=-200 80 amp=11
+  vowelout f1=0  f2=1000 -500 -350  f3=-300 80 rms=22
+  lengthmod 2
+  wave ustop/p_unasp
+endphoneme
+
+phoneme ph
+  vls blb stop
+  vowelin  f1=0  f2=1000 -50 -100  f3=-200 80 amp=11
+  vowelout f1=0  f2=1000 -500 -350  f3=-300 80 rms=22
+  lengthmod 2
+  wave ustop/p_
+endphoneme
+
+phoneme m //not clear before some vowel
+  vcd blb nasal
+  vowelout f1=2  f2=1000 -500 -350  f3=-200 80 brk
+  lengthmod 2
+  formants m/_m
+endphoneme
+
+phoneme f //not clear before some vowel
+  vls lbd frc
+  vowelout f1=0  f2=1000 -500 -350  f3=-200 80
+  lengthmod 2
+  wave ufric/f
+endphoneme
+
+phoneme t
+  vls alv stop
+  vowelin f1=0  f2=1700 -300 300  f3=-100 80
+  vowelout f1=0 f2=1700 -300 250  f3=-100 80  rms=20
+  lengthmod 2
+  wave  ustop/t_unasp
+endphoneme
+
+phoneme th
+  vls alv stop
+  vowelin f1=0  f2=1700 -300 300  f3=-100 80
+  vowelout f1=0 f2=1700 -300 250  f3=-100 80  rms=20
+  lengthmod 2
+  wave  ustop/t_
+endphoneme
+
+phoneme n
+  vcd alv nasal
+  vowelout f1=2 f2=1700 -300 250  f3=-100 80  rms=20 brk
+  length 30
+  lengthmod 2
+  formants n/_n
+  before _ n/n_
+  before (i) n/ni
+endphoneme
+
+phoneme l
+  liquid
+  vowelin f1=0  f2=2700 400 600  f3=300 80 rate len=20
+  lengthmod 2
+  formants l/_l
+  before (i) l/li
+endphoneme
+
+phoneme k
+  vls uvl stop
+  vowelin f1=1  f2=1700 0 200  f3=-300 80 f4
+  vowelout f1=1 f2=1700 0 200  f3=-300 80 f4 rms=30
+  lengthmod 2
+  wave ustop/k_unasp_
+endphoneme
+
+phoneme kh
+  vls uvl stop
+  vowelin f1=1  f2=1700 0 200  f3=-300 80 f4
+  vowelout f1=1 f2=1700 0 200  f3=-300 80 f4 rms=30
+  lengthmod 2
+  wave ustop/k_
+endphoneme
+
+phoneme tS;
+  vls pal afr sibilant palatal
+  vowelin f1=0  f2=2700 400 600  f3=300 80
+  lengthmod 2
+  wave ustop/tsh_pzd_unasp
+endphoneme
+
+phoneme tS;h
+  vls pal afr sibilant palatal
+  vowelin f1=0  f2=2700 400 600  f3=300 80
+  lengthmod 2
+  wave ustop/tsh_pzd
+endphoneme
+
+phoneme S;
+  vls pla frc sibilant
+  vowelin f1=0  f2=2700 400 600  f3=300 80
+  lengthmod 2
+  wave ufric/sh_pzd
+endphoneme
+
+phoneme ts
+  vls alv afr sibilant
+  vowelin f1=0  f2=1700 -300 300  f3=-100 80
+  vowelout f1=0 f2=1700 -300 250  f3=-100 80  rms=20
+  lengthmod 2
+  wave ustop/ts_unasp
+endphoneme
+
+phoneme tsh
+  vls alv afr sibilant
+  vowelin f1=0  f2=1700 -300 300  f3=-100 80
+  vowelout f1=0 f2=1700 -300 250  f3=-100 80  rms=20
+  lengthmod 2
+  wave ustop/ts
+endphoneme
+
+phoneme ts.
+  vls pla frc sibilant
+  vowelin  f1=0  f2=1800 -100 300  f3=-300 80
+  vowelout f1=0  f2=1800 -100 300  f3=-300 80
+  length 100
+  lengthmod 3
+  wave ustop/ts_rfx_unasp
+endphoneme
+
+phoneme ts.h
+  vls pla frc sibilant
+  vowelin  f1=0  f2=1800 -100 300  f3=-300 80
+  vowelout f1=0  f2=1800 -100 300  f3=-300 80
+  length 100
+  lengthmod 3
+  wave ustop/ts_rfx
+endphoneme
+
+phoneme j
+  liquid
+  palatal
+  length 40
+  formants j/_j
+  lengthmod 1
+endphoneme
+
+phoneme w
+  liquid
+  length 40
+  lengthmod 1
+  formants w/_w
+endphoneme
+
+//====================================================
+//  Vowels and Diphtnongs
+//====================================================
+
+phoneme N //should be more different from n
+  vcd vel nasal
+  length 10
+  formants NULL
+  before _ nn/nn_
+  after _ nn/_nn
+  lengthmod 1
+  after (i) nn/inn
+endphoneme
+
+phoneme a
+  vowel starttype (a) endtype (a)
+  length 250
+  formants vowel/a
+endphoneme
+
+phoneme A
+  vowel starttype (a) endtype (a)
+  length 250
+  formants vowel/aa
+  before N vowel/aa_2
+endphoneme
+
+phoneme ai
+  vowel starttype (a) endtype (i)
+  length 250
+  formants vdiph/ae
+endphoneme
+
+phoneme Au
+  vowel starttype (a) endtype (u)
+  length 250
+  formants vwl_zh/aau
+endphoneme
+
+phoneme @
+  vowel  starttype (@)  endtype (@)
+  length 230
+  formants vowel/3_2%120
+  before N vowel/@_bck
+endphoneme
+
+phoneme @r  //tmp use V3_r
+  vowel  starttype (@)  endtype (@)
+  length 250
+  formants vowelr/V3_r
+endphoneme
+
+phoneme E
+  vowel  starttype (e)  endtype (e)
+  length 250
+  formants vowel/ee_2
+endphoneme
+
+phoneme ei
+  vowel starttype (e) endtype (i)
+  length 250
+  formants vdiph/ei
+endphoneme
+
+phoneme i
+  vowel  starttype (i)  endtype (i)
+  length 250
+  formants vowel/i
+  before N vowel/ii_2
+endphoneme
+
+phoneme i[ //after ts tsh s
+  vowel  starttype (i)  endtype (i)
+  length 250
+  formants vwl_zh/i_dnt
+endphoneme
+
+phoneme i. //after ts. ts.h s. z.
+  vowel  starttype (i)  endtype (i)
+  length 250
+  formants vwl_zh/i_rfx
+endphoneme
+
+phoneme iA  //try more before N
+  vowel  starttype (i)  endtype (a)
+  length 250
+  formants vwl_zh/iaa
+endphoneme
+
+phoneme iAu
+  vowel  starttype (i)  endtype (u)
+  length 250
+  formants vwl_zh/iaau
+endphoneme
+
+phoneme iE
+  vowel  starttype (i)  endtype (e)
+  length 250
+  formants vwl_zh/iee
+endphoneme
+
+phoneme io
+  vowel starttype (i) endtype (o)
+  length 250
+  formants vdiph2/iioo
+endphoneme
+
+phoneme iou
+  vowel starttype (i) endtype (u)
+  length 250
+  formants vwl_zh/iou
+endphoneme
+
+phoneme o
+  vowel starttype (o) endtype (o)
+  length 250
+  formants vowel/o
+endphoneme
+
+phoneme o-
+  vowel  starttype (o)  endtype (o)
+  length 250
+  formants vowel/o-
+endphoneme
+
+phoneme ou
+  vowel starttype (o) endtype (u)
+  length 250
+  formants vdiph/8u
+endphoneme
+
+phoneme u //try more before N
+  vowel starttype (u) endtype (u)
+  length 250
+  formants vowel/u
+  before N vowel/o
+endphoneme
+
+phoneme uA
+  vowel starttype (u) endtype (a)
+  length 250
+  formants vwl_zh/uaa
+endphoneme
+
+phoneme ua
+  vowel starttype (u) endtype (a)
+  length 250
+  formants vdiph2/oa
+endphoneme
+
+phoneme uai
+  vowel starttype (u) endtype (i)
+  length 250
+  formants vwl_zh/uai
+endphoneme
+
+phoneme u@
+  vowel starttype (u) endtype (@)
+  length 250
+  formants vdiph2/o@
+endphoneme
+
+phoneme uei
+  vowel starttype (u) endtype (i)
+  length 250
+  formants vwl_zh/uei
+endphoneme
+
+phoneme uo
+  vowel starttype (u) endtype (o)
+  length 250
+  formants vwl_zh/uo
+endphoneme
+
+phoneme y  //try more before N
+  vowel  starttype (i)  endtype (u)
+  length 250
+  formants vowel/y
+  before N vdiph2/yu
+endphoneme
+
+phoneme y&
+  vowel  starttype (i)  endtype (e)
+  length 250
+  formants vwl_zh/y&
+endphoneme
+
+phoneme yE  //try more so not similar to iE
+  vowel  starttype (i)  endtype (e)
+  length 250
+  formants vwl_zh/yee
+endphoneme
+
+phoneme y@
+  vowel  starttype (i)  endtype (@)
+  length 250
+  formants vdiph2/y#@
+endphoneme
\ No newline at end of file
diff --git a/phsource/phonemes b/phsource/phonemes
index 2314841a..9b09cd8e 100644
--- a/phsource/phonemes
+++ b/phsource/phonemes
@@ -1172,6 +1172,9 @@ include ph_french_ca
 phonemetable hi base
 include ph_hindi
 
+phonemetable ta hi
+include ph_tamil
+
 phonemetable hu base
 include ph_hungarian
 
diff --git a/phsource/ufric/sh_sr.wav b/phsource/ufric/sh_sr.wav
index 3e7abc301ab40066cb38625ce2629ecc8dcb7d32..5c43a91e1b6799687cff946c784caad65ac7ff4e 100644
GIT binary patch
literal 5828
zcmWMqWpoq=1Ki!az1=<c%jM!ONU#t{kO09gK%iJ5NP{$OfwWjj3n|{B#R*c1HGH%M
zf)gZ=;+_z<yUV%D?%l0?y!kut&7YY!^JdDJkt6Ra0boY%^t`W%OJc+T0DzwnHyr>*
zTmk?H5Ce+}O9~Hv=KPOyg?LMB4$NWl1yg`iz8sE<mk6BnlzZC2b(jTt3?)!Szy<P0
zJT){F4009=`k`}umuxGDY|bq1w{QqN3Z4sQcr)p7oU`;oAOurL822{*w>BDw6b14!
zi0`TP`;j85+@8*=kg9l-Sh>C&yN!9v!|8O_0BWG(v@{G?Iy6)n@Pz2jYLecC3!Qs?
zT&NtF9Li%m)klP9x<2V|F@1${-tvGW&>>G2{nELjX#}q}wi!wC<OAn~=luB{67p+F
ze|RjKDDVX)xjx{{z;SLESc6~0?x63zD+vYtfo|dRg#{El*hZWoqC?k%LYOa}$zK}!
z(X~17HHbn-Xcgr|S4r0eO11aA@1*I9yVU$(F`7hDUB~f*%4Eq+-%H<3Y&N#kd*4~b
zD!?YOYLKbWVsCHXOez<vMYenS)ZFN5e#yrj&GN7_3AX}2n0c^If`sf=isA|~k-gx3
zaCoTIr(sgCtAxQW;jNEsK+IOe76*Xzd5R;85qt{fhY)f&e%sH5V&sFchENui75K_v
zbn$r8m2%;D3Z$|Eznbji6xAHK-R%Uo$Ug~x(%yQt9axcZSG^VJMgWc>{$KGyq4Qiw
z`G+9FE-*)-3ozWy!RsR=G2vcWqtaU`{tjN{?!}a97Rdgys&&soxuSA*ZEzfmB_GSV
z-afz(5|2`3FjuG;>@YLOvB0@B^cW6^r-ZFQg0??RRbB;aws@~}2omqvV~-0^&>wUz
zlq2gYX~O~AH{MM4P)-?{%U&i@QpJ`sq8rN!*9$j6C%rG2=bR7zX|{7P7;V71^xr%8
zC?5(YSd^wi$d{TaaJzE_JPmo|$nz|KCrD=rX!lP}F7%Gl;Hk_PLbvKFq_-FPCj?0f
zL?$9{=ry30or52+pQXMPhXlQ${nTBe-sdOYL90=mUJZ7Ioq`uOH5vXAyCQ~RH{Gpn
z7SsTxdwY8Im^|W{sE2)#&O_A;&a?NEVw!;}I3ua2zClbpoJ_AlmP^ZN($wF+3VJU5
z3poV`LB0E|lgGkQ3>XM3;H>5XoM+w#_8R{gax0Q7YZl(Xv#cqhTETI&f!adVfs=!b
z10r?2T+w1qrDL`W5U-VN$CGRW0`0;|-Ww{Gf&)vuS7<lyxnv42nuxO<C$=I(eb<cX
z!iMx=Sij~k^e;5NNLUZHAo3l_N@!-s6l0leSJ-ox#0^NxP@DaZtq%%E{tp^x*ai#|
zDP8MzS=jBwp3+HXwf4<toy%$sc&77>ccw%oZe|s@7TSBG6#^walOgda`id;To@^iF
z69Ti@f6%ee0@*aEm-_&<70PDP;fqX9hu?1yd?UCTB8fcy5aCWwgtmk*3mZ^2TPG=p
z@9Sf=``B-z7E6SV@xDCuCN}W4QQs0fzFQG6zbV%RNDr}B64yAR`Q0#-8rS)Q>5+1x
zdQLFQUn1DS-mGi1J{A2gOB2?jeD6|Ig>5A#AUZ)mMLJc_h}@Rh_Bbp_G?-s3x(+@x
zrMe0c3%rQTW~Rum^VaD;8qNw23XTWofX|RRrkyEf72s#B&qE6}Z<U7~8?0H#f1E0}
zpE${>;z4{EigvGbho}>98dOWI5}D+Do2S-;?}}{Hd_#$Yw_(0}g?=I{Nj{7F2$@N*
zby++Ig4u{2e#3#qT57lP9`TE$n7z|(2MVPwcsYK=w8UO7ZjT+zj?)!W10&vs{_B_!
z;EQc&H>xjl0+Z0CW(hf6!@&kSw0IMJ5PFR*p=8cplucYMcM_cjAJrx81gF~y$T)5f
z=s5ibnZarEXx#0{9LYiG4`-%-q96u7>|O*8m)EgGmRoiXwoy_L;QKxahoOv7=DsNt
z$xGe)EPW-_k&w4n{og(;7LskSiWrOZo&TBf4YWp*kG6-Rf`>q9@Ur_Udo6b^C_zqx
z8@+_f;)>?oO!$ud!AQ7wqYqR$5YTkTl%%SS@vxRy2fL?X!?`JeD$jDt%DK&3#W_Uu
z!;{&?fZVwc+8UJ&pK6P?7Ar4BA1BJRU(r_4BEU&Z4wkYD#Rc$e%L>Obel`DWU_~$m
zd5!jfmNGj+RqnW89kvhMOzV8-p)o?Qzt}n>DCS#b*=(EtF<lMHLy_!mpx;db)5KhA
zuk#)LhsY5%C^WH7=y?`x4y$pbH;iWnrQ?E?U647%iHo?%+H9x^EfDgsEv`uI6z=+@
zKOs{m;u_7Zk@f<^$p@k#Oigp8=bSW-9|f+!hH*=9Ugw+8DtRmV6Kf<f(>{X6!OOrT
zL0@bxUK&~q=dh24;$8QNN@+iSCw>ao!)Mut+?(y`f>Ko_Hy$Vn8N7>i1|y(Y+r5je
z?da=t3SY|70&MFmpj0v&4fn)Y@|YjE44l9NxqkZqvx{|9$`@_LuNZ^u|K!Vk{q@(O
z9tqpzdB#?A8s{AICU}z%2}a_@o`YaaFhq=yu%qS^x(3S4(x{a@*J$IvkVu)$+1It+
z{bNj)M%A(OQ+wpmzGCD2M{`5zeI2U0PRv!USdWrc0r8mslB;Bd*7LS+M3WU4f_L$U
z%n3%$(IbVv3i1Vdgs_DQkxR@#=4a++U=5}cuk-xrPG|0e?N~Om%(V|#|2a?9f;9eq
z=ZmiYiF(KI{M|bA+~1S1K>3R{pTD<9b6>YXyO>|1xZquG9t)-kQju50Y#J5~WzX-*
zxA~=hq1v?;vMZJ%`Hlf56LUd)i|Q39;x=Gqfg9c;@-PIm-h)HHa^_$7CI|<%2GW5#
zt{UC!UFcoT(TL3)z~-?r!U*v&T;TCC4qh{2wF~HpVi1@MsOYopmE_)p4hfsS=(My9
zquZh{iA2HPPPRwMjYRkOFF|=^xHn$)tIFZ%ZfRw5P<i0z;1oe7Pv<|3*TQY!^N^2?
zGZ91$w+Jq_I($#z_mVlPa%NjQ$Mm-}EMiAMs!t>5sN=aI%R(n0K8|d3n1Kt5B+f@1
zH5XcMiZ(^wL({#D_FU?)a0$6nYX&pZa#dB@Ki-Wb4#qu;*aR%G+g;mve@c6JcQi_I
zUHXK$G4{DFJ?K#~JLh~b*<I<>3SFw4KyRIuSrzsPd**s;<FI`E^-MO9!v7RNOj_!P
zxGnsv_076Rs$o%b#^cTidf>zCqqG^%0>&caLfh~=(q?(E>(r-1m?q6F{%D?Naq<R0
zhuY@TwNYdEzq!UawklS~C|Y*kyMS;<4~p7nJ7hd)uJ8jazHqSQtl&CW;UD9|@K}62
zuz<Hf+9IqC%+jHz8_2w*W0LJ%(S}@1BRb$&WgUS|joym<Wn36|!s7xv$v^y;s2_zl
zFfwq+G{ApWIg%UWbkg$`(R{V>A7jPmUHG1!$ms^ZfG2?mTs1~Jcq-;QNe)qCSwgWC
z#jf^ud(d6o7fQdOXX{QIl<Fs<LhUKMoAg&%HTt4;enUpsuCz^|J*HNkP&30Y{#7?<
za@NnhOHXQDLvw}UqUKS$A<zINOR&Sq_pv#DdrQ1R1;Jes$`G2M9lB!gV?`PAM7P6a
zlU_@Fi>%f@c8n1nME8*~oN(Ceae4DN#h?uTo%bW}1OfUU(C@M9tS7$ftk3SlpV40m
zlknkPQA~NHRr9B{uGs@B!oTF$J*nQYXawx_E=E#0k3GAYbYV}{_~5_pC!D?FsrV?5
zinkmq!W->>durs*Be#;zY`M4}ABd8?N5D4vFCP{tVGaR}sy^cH>Sq}5CAd^abuXLd
z2`>)3#=iH6)RiSaNfQ~?K8=>v4?h-q_(bElma0kjv<hF0k9*&vqWwYrMNV<TBI$41
zX49*P{iwEm1{91+1s+)T+giynVjQ`RGzi*s4_f<6qazW<?7i=(hK6V|!3v!of2QC|
zNl%mhA~8-qQ9Yb^tY1NBB_G6Tbb)rRIX`@5ul&%G_xWylOsc59eVIcM@tBw6c@Fer
zfzHV`fNP2Sn%vSp0r^@|>Z<W<<L^^v!Ffh+=P$gHa3%8#bCc_|wCdj@66F~|H0Llr
z!ukPNhE|X#103Ky{|;uhXPM?=h3aG6J%Lv18h<i(8uvPZ*w?c!B+d=Ry_`vu^@~$-
z+nyM6H3F`x?nr2BpUVPP2ie(^ycQiNobTI#lpr@8Medp89r~^OlQhqA(j=B!Sd%O-
z!F@tM_*-Z={*bj0&ZMpIVxHD}ySao`q^pHe%Tf13&UQHH+3B7azEFL&qp1Bi$=Jw`
zf->K_hDP_h=-tYn98b-5s4F@?+|ac4&6m>SJuT>9!%2O2ZbaBQvcGZY=Zz8>wVL;j
zzb}v#T-*LFaX=~*Rfy}kw1LuBgJYF@!*huDpA~OPvpqYQw6J;H8`M(MmCmW?GK~^t
zferLD&k;(6)!-}rTJ#^mi9n*^3_LQa4T-Ibv3#w}RF*@l+=rb7ta<X&h|oUVdjoA@
zH~Tl>V>#8*5;)Hhu>Zgt$9`a4${XJ8FF;s#z;aglj_Y%EXA{!P%zFGc=s5=ma?N=T
zCM-XE9(2w@w(SmWNd7|bu<epFN^u5@a$e+I;ZAjI^pBQlk?HPIYinS+xKr2$CZXrS
zd-krb1N4#T<LF-BS?oHpMX%B~3pc5|f*lADUa2)bFA^DZ7x2N?xh<t?Ahy((VO+{#
ziK2ipR4Vn#TJ8IvdcWq4@2p`<aF+C6FsAKQ;9^ggaFMngmvVMdcl_m{`HDMIsWaIu
zV^37Jl1Vz72MNnnz(lrdgXod?s3+6D+WmF#WK5jmSbK}<g2baN1mCn@)(`B?$@<gQ
z|IU6l-E+Jm*!0{}BioEZ9;ug>{39*XzYB~?F7)v3<avgVl=kgacjZ5Q<hKI~j0c||
zhOcBM!Ry~{aS-7X<SYwEH#Kr$Y;jY#YgeB>%44tiZ?hz|-G1>8u+!3Q=w;(6-&H}G
zcq*jv>sg;T58Hoktdh^~M+!GLLhUc*f~0@^uNqUxb7^U^l&0TXg1mFl^I5mtgTbA^
z6mN(L6OX};lfyl*)6OiH3c+38bhL<9>y5N^vh+POB{%*fKMssb>DisQTfY%3OWO?O
zyk4hYBHI*KrG5|?p!JzQ3Ia%%#Yr#Dz$Iti$2a{G-WL7Fm|~G9E>!>7(7k?w;!cl=
z?CjUGO!?jE<X6ob-~J<QiT2`7Pom%~zsXzFl^x`YCdu}TCwNlK|1t6WIiZtQ2iP9Q
zmJD_8G~aa3=V+9$FvqdPvrg<`O*QwSRpN5-P~VysE0i(#G-`NIS?`m->eff`RmWqK
zK-0?mx@(mOQWdKY1s*l(UA#n>P~6#%v@1$@2VBMN|KhnR1EcGWt?k9=6J?dT=f?v6
z_91rRs8`EuKP7HwuWh&*_&5Gx)Gh+58Lc;Gk_l&8dYUNlG35tnqV|C^F=mWvo8y3e
zs3S8JrmT%z%9ea6Z{{UOCE@MAoA|0Q#blz7Rv)5Nae|=Ze%-4;_2@@&yKlukU7B@n
zRAh(ae1F!&NnfE4p0#`PQ<o`kY5Uc`kPOS%?|WHi#rnm>;I7tD6qZUz<qZ!zW#R_q
zEXf`p*{TMmVOhZfM<l;l-3#1nc+{E!UP~M%d){@wfsa};HevTZ1p_xz<<#n$mDGs=
zOO(@Ex*Rh^t3z{KUx-}tZ{67)h1@&6{s#$PR=RUybCBzuBOEWqqmo<EI_)*jkD+f|
zi%C#^E&3uexG68NOr3!~_J44H?`{q>2wwIuhPAxcJod2$4V)R4{br3}Saw9rqna(R
zdnVi-a+Drey+;>F7_FS+0Ng)GQY1ZXS=v$HUt#mqhpo>VP+p(-$k6kaQ+_yNW6CKo
z)!G|4;ePZ!(myIWBd%A_uU!pQMX$w89V=PKB3rqWv{=`A87ONEen+eV0M^gW4W5mu
zKPRy3zJI$|B99qErrJG%e(^=N@wZt1bHhZj)wRavfvTd6{mk{}oyPX~*NRtxEf|}4
z(=J1l;lF{Xah<j{GOXWLo2a1*J`}zHifF3-^kp>McLQ?U+GXj+>SEp1U7&Son-wGU
zQ{VeNhtj(u;!WbtQOp7L6L^KILYWcX*wCTfqWKncwKXxn_N`KV{a@<G6A42S@+=bV
zud=Sht3Kod;Qk?OePlYd%y1VNqdp=iu<!4@%AX>S|MYh&khmnL1)cfW^z_d#_mJ0I
zbxWB+EsK-=YdD5)O?kz?+(7Cg6e%&6;Xy<ad5)N4D{_B?uSya{@z75`QSc0N#a~Kx
z(A$NVBqrZ4R+q-YwLA$kZ^=25)~dbvdYkXQRL{9Ay07})7t?ebc-><QHvGZruI_`6
zYvwlRciom;NlK(@Ki&ydsd1{(B&5xfV#E|=nIL9H9F2gjyNUjqAC+_P(xz~0e#(%@
zhNi5}ovOc+3Xpdlhb&svUf$`hW1bZC_weABHtXG(w6H(Cr(Hh~AjT6YLid?Mt^uE{
z^E&Fpr=xcB#+g8JR{RLKcl}J=R#|$YlD)9Y75Ea357h;NEH6KW(Gntd3Gl&PNGFE1
z^1}=#41Ia!-08v7ypLj|Z&FjNe^K<4@P0v?E*<|6=^|^}Am)y|i}gid4q=oJmsIJ}
z+g}Qfrp$vkv{v}8aU+o;bbMeU(;-h~zVzLo@_bMH_px|>BE83$fuD|z73Y}-cm@e}
zDAb=DcPe{u1OfH5?{O?gx5$I^J>!hv5amhb56r<X#x$QF75<FY8UOZ|DVv1j%r6~P
z%E<5p!lj$$AE(a2tcGS=8RFDbVU_x9*Yfaud5yi?F@(qGM%phreplX8mK%F?^p%cF
z4rli0ihVlPwa<T*m8^tsBlnmpVi#G3tYlpRZ?cO*>%41mC-<^E0jzK+agdn6$yR+t
zCiwm$xX?!Sv>@nT!tKDuSt>oHk^<ou<Vqg|48#a>y=Mk6LfFb)<5`E#1oJtwuwjAs
z<W{mVR3UhRY5ki#nW10M-4YMuXsaaWCB8!TSXx7ikZ8u|{@!&#u{PqReXIjT7c&!Z
zCb%47!eiJ=9DS(Ap;!ygw^4dDsuTU$JK51`d=OX`CReeLZJvGlKK>YCI6sj)9ovIf
zThx|;?6px<(qH|2Tb=71nTwTTmEcl#Z}#`(K2{6VBUm4r9Na?wh_Qfh|3v6C_d8ZG
zH95GGKEPUoWwJ(jDjdVfc0R^4(bHWIJq5@P`9J*YRGGUpWMGPkBBno^L~diO;BntR
z8Uhn|jc{1-D%{A#I!=3gbIzjy)*Ij~HIw>^971<uy^!jlg}nngZ8&Nu5q+oXWXS_7
z1E*Mvfdl?+?0e`YCLKCM^6=A)A6A1sSW{W!aF_2l=oo7T6vv)SB~Uw92bsaN9-zTY
T&MXc$cnJTNxlC^YHiG{Lh2u54

literal 4652
zcmWkycU)HG-@fm&&vVW**&xo+pAinsk!2bwYT`~o9EI}Y0R6cwQFCSGAWbQAq?q_-
zPSo56q7rDiFn6MetcQnZo#*WPoY(jB{rCFk`r}&PF+)>Q-f#hs_Tl(J*|Qh6bpQb1
z>uWzAfPserz<>i}WiHGFuQ9nG86bRwF#u(=t;q$#qTo~G0XqZ_*9)*La;=&ld>a<K
za!lIuzS53qjeUjsF-`dJ0HHr-)1*>mw5to_Y<Bt-=Skb1rldgY7^m${;{f5ZeGItn
zi?t@k;_klB-?JZy_|b={EfIq?SKU1GoC{I6m9yx)_;H||bHy;`nFg9Hwa)kUmik9u
z;&0dS%_VFEv(X&Jv|-lErcZ%OY?Ha3YJum4+2rhI<OV*%vzam4Qhgb*mH5**um4Kk
zgSG0rdbR%0e2$&OhGB1mFf>I=(h6avWgyeZ_(vHjJrEK|CTu_RgN8Lt5B7I1b(9HR
zlsfAnP};1iLtIyFC-kO<QmG{JIv;Gh>TTm(#8w1~L5cN&w#L&NOLg6c)4VlsqxCm*
zNK8RI`7bpFO_oPvtdWcf1WR_-hU>!_ATJe1*v?p~P`ns{#e;9iWAs2dN3&8djc8=S
zUeMJrHn32c#7|-WRYs|+;Yl;f{12+bhZ6g>d*BY-0WQb~%<I;3YOXtmnD6LnK5i=2
zuf^JkP0imC|F!%mWryroDZbI%iWg$Nw6{?!kOnd+jMx=CBYnko=Ok&pe*!+yR%_mu
zHsS-}XxN{;8mf`YxVNaj`dUz^x|Oc@32qd<UTrU?%gJ~>_o+TZinUe~OB+ua{bQCG
z34!aRKxO)-E2kq?+VqzC#^?MI$8z~dNMbUqx4lE%A6UbkgxaXRg<n+`;eRruGFKRk
zZDa0Zot1y}gH#(VN9zVolb^%4%{X*kn}O2cBjTWWT>Bf#C5!Z-`YZ4=zE{!oGqxp$
z&EFRP(SFIOcjw#F!jqpdo~G89EQw9qutdJIe}oVfnM+8undDRF1#Nub5IN3q(Hma3
zEc|I)if4E8cxR$xLemBRn20>`Nw7Eh4o!vbOVjcFmH;`%tirQVZp$e1QrJ8qM*3Ze
z(`9Oo4au1;7TXaXc*n}Et+~XF&^g*qCIrqIH!KNqq`cOtlE*xPG$-n;_0FsP?$fSh
zdr7ESNvGaWbKq}$XLwOb1^MbJAxM7d7)JKgr)iErp;;WZ7M1#*lPQ)|{egZ8_AoQd
zY}AT;h!0ZcX&EL$2eH0%2B=04j6J}s_0hkuO~o#FM7$@z!S4>8<g2)(pha5YtYfbD
zLP`<42W=5L(a(8ROKka$Y_i^vIXRE{7R%9#i6?q{?E^3jD<WsXP4arQ*U}b~JwM@E
zRE1)HIfR`XbH($KTaP##8C`d;_GmjSEUubR-idRewEDvw7yX&H_+`9(YP2Ek^=zS+
zIwsX^_jYXmeR!XmTW&5o#`=%HOgzTdv21fsZ`<e#j_u7h#Z4tDlko%LXW_p3Dy_yb
zgcvG5MW0i@VvT{BW_`qbGRe2iAIDtewy1rzwiY)S+cFHZIuuz9_O$n~&S<Llr*K<s
zFN_t!IqEYaMwmz75kMsT365>BbA4|!GxATu40cih_ZIwHnaHL&p9Oz-JfBwEtPhoY
z6XSMp#Z}&xhPOV6nOjY|v!X7O7n)FLV#GyjO>n399Y`lDVV1v8-4K=m7i)i8A0wNm
z9Q7jio2R(tY#*_+)L9*)te^)*RdbmQzy7<6^hf@MEisyHne4?UFPi$aJLUS&JvPuC
zuh9FE7x?yc8>LD+3x1ID#B$pa2MS2uZ-{o>MzO6lJ?aB%g+E`uh(*A$YPG)1ei$$C
z=P*y`ZOyB-blXO=R{xPISDy+O*=^QpVZ0t>fi_ZX<c2Uef|;<Zl|VI8p>_lfAdZ^f
z>tosRM7lUk$)LK@L89L56}m1S$L`zDkaA$8dYamYT!B}Xht3jttTB!};omF9aFg*k
zsDLS|AL{y!&{~>ybW)BBZ}MI&&OH?z37do+Xg;NtbJZGa<T>v`t2;@C+zz%as+YUN
zu6Ql_M;%QL;?IKp{=yJzn3k>f8f#H-P4##<B(fvz(aI5^M#)`#Pub<xL+TXqq*-ci
z)>`0N_6V6Pm78fyoj6_zca6s%yT3&LiI`x<HV3hf9TOzf9B`eD8|&NotTU{PXf)Qy
zHcKCJci^s?L+bckY=b#gn_%uB8RaKou01YhmN>Ak*j{OUS+}tHb{uf+YdS1`&2wO*
z+@GJrEf4tv@0s(cX9Q>Fg7@hD(sp-<T_694n$x(87RUp>XgJb!7z+;+f#G~|V3&89
zqd9D@y0GaKR%O}dnJk`s^R}bSliY^i?TU4sdK9m)Ofp*noc0Z!L6qs8<!@ms)0>Ia
zBE4?yq|0yZq8x1TXc*C;+u>?<8#PTY3uPFs*^&5Ba2VIMhx%-6Hpn)su~*dY&_-_o
z<&A7jwDY8eI_jlZJI563P_SAvHBN*IgI{9arasX>RU=S77=srwd)Y^*(PQ&>r=ob%
zGK-H<pVqzaxAVPXDv2SU-R=Ztmt&7|Ov7!I-l4gazZ8B?c+r4c=2&M1_kqjK{}I@l
zZ=5gNCJ6Ih)nE_#{$MV;OGRNH`)7$)XoVU88>pMa=s;z#nCWZHH?}K@#uZ|_xm2df
zh1><@ub_=)=@z*jr2-r5Ctd+bQFCFVdrz<v`w?%*@9Vc6iDZ%Irr4e=GE3oT+YOBM
zoHM_+-}K#;|BCqm9&Ol(&*!@eANf`>^CJ5ZE$%OTzmU%?^Im7wDf8kh&@wS{ePp`8
zhXA*bXz+w&OZY%~w&$3U819h{G)8lb^E-8_@IDdGt<t{^xV4?GXOTr>pI1YnXg@DH
zrP*PMqql!h<69K4&CvJbpD>%H*}?;A9QAXEH&e-pIKe)KJ%jPVr_31JS-qb$jV7p+
z;1+3}rK4>qye*{ZB%B$F2R&`W==sVn{R2=GjMn?uPut%^<x-xpL*5vAh_ADjU`^6D
z)CToTV=T2d<}~%Y%&I9$nC39ga;<F3Q9C)`=&7cud#OxIik=&oOcn84Jp)>rB4We8
zmA1DeNgJp{cDC6S-;QocpWyGZ-=P=kpTvG_b|_xi;8@Ro7A(~?VzRzQFK4^ql=cep
z*zejgV-$G;&(Rm4hbT_Vvy7vs`Y(!unfHiXk;m#>`{k;JQOrsEL&Y=-=<aBQ*h3vh
zA@Zs|9G^x#Q2PZg!CFT+tX4--1m^P6!b2{P+pe@RRuG#6Lx>Ff(Y`$}Syrr{B1>aC
z>VD*C9C~ksX7P{NVahXYAHH8+qC9hCv5P{bVk$O*SS5zjMXrM~+x&$>+4J}X;!9(q
z?R%tpG|<;M6V`dk)LFz(vyeE;^d?i`GG$lbh(3iy#3SS;bI@YXOf=Pb(5f32EgU_N
z+h`X6VtcTu!j&LPZ^9!~0uHwPfxQ{(CFK(9$kytoD$hiRg@n$v$MK{v7scwWuz2l9
zpMt(}R<Y^g7G)a@KrgD~rdzs51%cbj-*^Z7B{RylS=wDc7Co}xCidwExMIAze!Z`=
zZK4aIV|p)+*MF)n(rUsI=xV7P9Ab4XK6F-(ps#V=u@WIM^fOUR-!z)hGV+jd22I5(
zj2%Rh;Sn+P3)2O&OOL6Mq|oe@4@9qFYMMeywDo6T7nXvTw!LPvcM<90_XN6|KKpKI
zn7<7dZ>^H5Lhq3E^fBZG$@nQPsBgg|uqRMMN%~{5%DgO|&_BRaurT6R?5<uZ&B9{s
zTg<uskIeI=S8flkGnGc3u}WPW3PIgbzz-0T0~g5;$qz&GUN`bem2wZk+gKhcg&1oI
z!#vs*^=tJ_WZ*uwHT7<2mH!@GiT_1ecuY(8UBTb8pI3kOwxSvwIm#9PJ?@rsX0Us0
zIkq)ynVF(rC3^>xgQ~SVdq|Cd+wo0^Qoq*9;8$2S<t2v7vz2Bz53Mtk*&BR2Gb)s%
zy3smg40+e8<A3`e1YTe?wTNto&m}5Qn@}#Xl)kOxU^#qt=(qambVh`!7{LYnFRaz`
zA~-mvEB{5)3Mt9DS}j&D#WnF)pM2bq=;~`-4i^wf<YxU@Gt`Rh3+NA22G2L~;2ph#
zC6kznYUvwbv!7Pn++52juwUhr9P=bNZ}!3}m^IKLW-B6@fhQO)W-h)%=qCa?ntK~t
zh8pE3!DwTOeTik0JXC!_lxhQ-^ReW(Q`kGcKeR045!weXP>1aQOTF}U^bG?l5l@to
zub5BNK;u0$Epipt&0i}1hd*SNd5W}(u+O+UPl8-&lh7(3qAOfmrJ#QbnNQx*J7|6|
zo+)Rp$ou@KF`K0f%@HBBmVKa{4n;x-xeF`==czeB5YNDJ@_VraMT9LfdbMQX3b#rt
zl+Mz#FrRphoWs=nkBTA3I=YAdxl+eoCzh&jpdHj~Vv%6iPB^FGy<39jQhRUR<)34*
zx-N@u?{RD++XF;l+o)CY2;UPRTbI1vb#Cxn`%-?1CR32j#d6U^rAjdL8DX>7aL;l5
z7rtC$r8F|1vEl(GR*VO`NgRF%-S{;lOF54Ef>=D3&LaAt`{GU6t*@kGZQm0egFS`I
zXaNz&Ol7;nQ{J^wH}Wi)Pfg@kX%!x~-j~Hd8H%zL5fgob<yt<C+ARi**Zaqa*8U(~
znQ_6XV1si9wbSR3dRo6Deh($<b;Lw+6f9H*=(p&_RA*xhC@_$k1Yd&LiXH2~{%Q7B
z;$bTs8t17jX1QdScM<)~_NW2hN~EIA`e2Y?Jcg-kBw2_YQn~mJ^D9#jx~oyPR#b|x
z8V$2d&`EIu{fy}$|G(<<QKe8q%%{%;Ddn=QKet}3QO@Z*)i5)LnNH^^rvEa?VgS*J
zd<K3ELU|G)(w|TvbXQ`OVzY#J0J~wCq=Y)E;TnYxrve~P0Y)Y3!Lp^3XepJi*^E}q
zm$;#KG&$^8=!N`<IZl?rNno6GQVoz3i4L&EL|`$Qhjl>VN`>Uc58Ap=QA!v2ikU_J
zmsl&Wlo#5{xK3WKr9bUv`k{lMJ-${Q_?q(_^O~`MxP~#tHNAr=QR!HW!lE<OaH0Xt
zH+sWg^scYn`>u01CiqsXRoG{68EmB9Ay#Oz(QL%1=|qV|MW4&v%nrmsd97~{<F$XP
z9u#u0EGE@>A?!jUm}~e2qYQJPZN8qUfiJc)_yDPgvP_=}uToj~Of^cshwm^?DWky#
zy1kJu#u6zMt$z!DW3H<Ay-ST_{7w>SuZ-E)SS>;PoSf>gkpK9P2l_*T`ib6-O;XOu
z*NtrR4SW~%jhU-#(QA$OK_uRn>H@ba<J4u?aiUJYVn#E+nJp4)`iS50klEfWfVaR3
z^AF-amVq{F<Fpy(KKd_y2B-}h(n?dq*3fs^QUi$Bv>)JMJPh=~ejw_N8tuA~fNWYh
zeur5IN1+L?zX8itgMMOR%*N16Js$fBJ;F<@KjDJ}Ub=-Hp^g%N<9SA0@PEazOaY}D
zGl5`ytz;`($W+^O+$R?JRpk+#PiMj`>?ZtP8m!!*PSaoMNpKKVj=uEU&|)jkY}7hx
zEAatj43Tb5g1eQWI?22XF3RiSIAWZ#R*d5A*jrGUxKb-Xi`B0{fu$3@M;fi3P@gE_
zbOBdl;PPLX6WkPEfa9!BZ!N7dlJWMi5Z<Cj!NWp?G!y>J&Zmaxz2q>dO#d(SKGO}~
qWUN;|QnR#cu!6~>#~2wxwfuna;E>+WWtr)LFfGy|Q4fs`;Qs)Gy7J8c

diff --git a/phsource/vowel/a#_2 b/phsource/vowel/a#_2
index c163a58c1c24555ede5a6a7c08651570ca662a44..30cb011aafc63642c731d4542afcaa62700b161b 100644
GIT binary patch
delta 113
zcmaE0@Ww(gIKb68B-k~Og@J(~Q90g-eWRc~3!~TOK$fq}%smVYn>{(cFiZS?GFt(t
zfPuk*ZOyzL5Ejb`2BwLI!i-UyV>uo$v;3Xm+%VaZN1f4eb0!anKigu<W?umg7M8#B
MjOsQ=im<Z)05EqUsQ>@~

delta 78
zcmaE3@xVYZIKb68B-k~Og@J(~Q90g-ZKI$)3!}s4K$fq}%*X%#-|Wfpg_-5gDTUdS
ieFfARZ8nz*JOJ`9a5Ha?6yacD`E&K(^v$Uf>?{BUQ5$&x

diff --git a/src/dictionary.cpp b/src/dictionary.cpp
index 932f0c32..2f15c184 100755
--- a/src/dictionary.cpp
+++ b/src/dictionary.cpp
@@ -976,7 +976,7 @@ void Translator::SetWordStress(char *output, unsigned int dictionary_flags, int
 	int done;
 
 	unsigned char vowel_stress[N_WORD_PHONEMES/2];
-	char syllable_type[N_WORD_PHONEMES/2];
+	char syllable_weight[N_WORD_PHONEMES/2];
 	unsigned char phonetic[N_WORD_PHONEMES];
 
 	static char consonant_types[16] = {0,0,0,1,1,1,1,1,1,1,0,0,0,0,0,0};
@@ -1040,7 +1040,7 @@ void Translator::SetWordStress(char *output, unsigned int dictionary_flags, int
 				// followed by two consonants, a long consonant, or consonant and end-of-word
 				weight++;
 			}
-			syllable_type[ix] = weight;
+			syllable_weight[ix] = weight;
 			ix++;
 		}
 	}
@@ -1164,16 +1164,14 @@ void Translator::SetWordStress(char *output, unsigned int dictionary_flags, int
 		{
 			int wt;
 			int max_weight = -1;
-			int min_weight = 10;
 			int prev_stressed;
 
-			for(ix = 1; ix < vowel_count; ix++)
+			// find the heaviest syllable, excluding the final syllable
+			for(ix = 1; ix < (vowel_count-1); ix++)
 			{
 				if(vowel_stress[ix] == 0)
 				{
-					if((wt = syllable_type[ix]) < min_weight)
-						min_weight = wt;
-					if(wt >= max_weight)
+					if((wt = syllable_weight[ix]) >= max_weight)
 					{
 						max_weight = wt;
 						prev_stressed = stressed_syllable;
@@ -1181,14 +1179,19 @@ void Translator::SetWordStress(char *output, unsigned int dictionary_flags, int
 					}
 				}
 			}
-			if(max_weight > min_weight)
+
+			if((syllable_weight[vowel_count-1] == 2) &&  (max_weight< 2))
 			{
-				// different weights, don't stress the final syllable
-				if((stressed_syllable == vowel_count-1) && (syllable_type[vowel_count-2] > 0))
-				{
-					stressed_syllable = vowel_count-2;
-				}
+				// the only double=heavy syllable is the final syllable, so stress this
+				stressed_syllable = vowel_count-1;
 			}
+			else
+			if(max_weight <= 0)
+			{
+				// all syllables, exclusing the last, are light. Stress the first syllable
+				stressed_syllable = 1;
+			}
+
 			vowel_stress[stressed_syllable] = 4;
 			max_stress = 4;
 		}
@@ -1253,7 +1256,7 @@ void Translator::SetWordStress(char *output, unsigned int dictionary_flags, int
 				if((stress == 3) && (langopts.stress_flags & 0x20))
 					continue;      // don't use secondary stress
 
-				if((v > 1) && (langopts.stress_flags & 0x40) && (syllable_type[v]==0) && (syllable_type[v+1]>0))
+				if((v > 1) && (langopts.stress_flags & 0x40) && (syllable_weight[v]==0) && (syllable_weight[v+1]>0))
 				{
 					// don't put secondary stress on a light syllable which is followed by a heavy syllable
 					continue;
diff --git a/src/intonation.cpp b/src/intonation.cpp
index 06c4f18d..29beff1d 100755
--- a/src/intonation.cpp
+++ b/src/intonation.cpp
@@ -43,7 +43,7 @@ typedef struct {
 	short pitch2;
 } SYLLABLE;
 
-SYLLABLE syllable_tab[N_PHONEME_LIST];
+SYLLABLE *syllable_tab;
 
 
 static int tone_pitch_env;    /* used to return pitch envelope */
@@ -242,7 +242,7 @@ static int drops_0[8] = {0x400,0x400,0x700,0x700,0x700,0xa00,0x1800,0x0e00};
 static short oflow[] = {0, 20, 12, 4, 0};
 static short oflow_emf[] = {5, 24, 15, 10, 5};
 static short oflow_less[] = {1, 17, 10, 5, 1};
-static short back_emf[] = {36, 32, 0};
+static short back_emf[] = {35, 32, 0};
 
 typedef struct {
 	unsigned char pitch_env0;     /* pitch envelope, tonic syllable at end */
@@ -285,7 +285,7 @@ static TONE_TABLE tone_table[N_TONE_TABLE] = {
    20, 25,   34, 20,  drops_0, 3, 3,   5, oflow, NULL, 15, 29, 0},
 
    {PITCHfall, 41, 4,  PITCHfall, 41, 27,              // exclamation
-   20, 25,   34, 24,  drops_0, 3, 4,   5, oflow_emf, back_emf, 18, 5, 0},
+   20, 25,   34, 24,  drops_0, 3, 4,   5, oflow_emf, back_emf, 16, 5, 0},
 
    {PITCHfall, 38, 2,  PITCHfall, 42, 30,              // statement, emphatic
    20, 25,   34, 22,  drops_0, 3, 3,   5, oflow, NULL, 15, 5, 0},
@@ -376,7 +376,7 @@ static void count_pitch_vowels(int start, int end, int clause_end)
 
 	if(no_tonic)
 	{
-		tone_posn = tone_posn2 = end-1;
+		tone_posn = tone_posn2 = end;  // next position after the end of the truncated clause
 	}
 	else
 	if(last_primary >= 0)
@@ -880,6 +880,10 @@ void Translator::CalcPitches(int clause_type)
 	int count_primary;
 	int ph_end=n_phoneme_list;
 
+	SYLLABLE syllable_tab2[N_PHONEME_LIST];
+
+	syllable_tab = syllable_tab2;   // don't use permanent storage. it's only needed during the call of CalcPitches()
+
 	if(langopts.intonation == 1)
 	{
 		CalcPitches_Tone(clause_type);
@@ -892,7 +896,7 @@ void Translator::CalcPitches(int clause_type)
 		option = 0;
 
 	group_tone_emph = group_tone = punct_to_tone[option][clause_type]; 
-	group_tone_emph = punct_to_tone[option][4];   // emphatic form of statement
+	group_tone_emph = punct_to_tone[option][5];   // emphatic form of statement
 
 	if(clause_type == 4)
 		no_tonic = 1;       /* incomplete clause, used for abbreviations such as Mr. Dr. Mrs. */
@@ -907,6 +911,7 @@ void Translator::CalcPitches(int clause_type)
 		if(p->synthflags & SFLAG_SYLLABLE)
 		{
 			syllable_tab[n_st].flags = 0;
+			syllable_tab[n_st].env = PITCHfall;
 			syllable_tab[n_st++].stress = p->tone;  // stress level
 
 			if(p->tone >= 4)
diff --git a/src/readclause.cpp b/src/readclause.cpp
index 44a836d9..73e0e432 100644
--- a/src/readclause.cpp
+++ b/src/readclause.cpp
@@ -1967,22 +1967,15 @@ if(option_ssml) parag=1;
 		}
 		ix += utf8_out(c1,&buf[ix]);    //	buf[ix++] = c1;
 
-		if(!iswalnum(c1) && (ix > (n_buf-20)))
+		if(((ix > (n_buf-20)) && !IsAlpha(c1) && !iswdigit(c1))  ||  (ix >= (n_buf-2)))
 		{
 			// clause too long, getting near end of buffer, so break here
+			// try to break at a word boundary (unless we actually reach the end of buffer).
 			buf[ix] = ' ';
 			buf[ix+1] = 0;
 			UngetC(c2);
 			return(CLAUSE_NONE);
 		}
-		if(ix >= (n_buf-2))
-		{
-			// reached end of buffer, must break now
-			buf[n_buf-2] = ' ';
-			buf[n_buf-1] = 0;
-			UngetC(c2);
-			return(CLAUSE_NONE);
-		}
 	}
 	buf[ix] = ' ';
 	buf[ix+1] = 0;
diff --git a/src/synthdata.cpp b/src/synthdata.cpp
index 693ec546..522f4910 100755
--- a/src/synthdata.cpp
+++ b/src/synthdata.cpp
@@ -35,7 +35,7 @@
 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.29.26  18.Dec.07";
+const char *version_string = "1.29.29  20.Dec.07";
 const int version_phdata  = 0x012924;
 
 int option_device_number = -1;
diff --git a/src/synthesize.cpp b/src/synthesize.cpp
index 680c836e..c6ca9fb6 100755
--- a/src/synthesize.cpp
+++ b/src/synthesize.cpp
@@ -33,9 +33,6 @@
 #include "voice.h"
 #include "translate.h"
 
-#define    PITCHfall   0
-#define    PITCHrise   1
-
 
 extern FILE *f_log;
 static void SmoothSpect(void);
diff --git a/src/synthesize.h b/src/synthesize.h
index c08383e5..00c73b6e 100755
--- a/src/synthesize.h
+++ b/src/synthesize.h
@@ -18,7 +18,7 @@
  ***************************************************************************/
 
 
-#define N_PHONEME_LIST  700    // enough for source[] full of text, else it will truncate
+#define N_PHONEME_LIST  1000    // enough for source[N_TR_SOURCE] full of text, else it will truncate
 
 #define MAX_HARMONIC  400           // 400 * 50Hz = 20 kHz, more than enough
 #define N_SEQ_FRAMES   25           // max frames in a spectrum sequence (real max is ablut 8)
diff --git a/src/tr_english.cpp b/src/tr_english.cpp
index 6818d584..4d48a4cd 100644
--- a/src/tr_english.cpp
+++ b/src/tr_english.cpp
@@ -43,6 +43,7 @@ Translator_English::Translator_English() : Translator()
 	langopts.stress_rule = 0;
 
 	langopts.numbers = 0x41 + NUM_ROMAN;
+	langopts.param[LOPT_COMBINE_WORDS] = 5;       // allow "mc" to cmbine with the following word
 }
 
 
@@ -52,7 +53,7 @@ static unsigned char initials_bitmap[86] = {
  0x20, 0x24, 0x20, 0x80, 0x10, 0x00, 0x00, 0x00,
  0x00, 0x28, 0x08, 0x00, 0x88, 0x22, 0x04, 0x00,  // 16
  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x88, 0x22, 0x04, 0x00, 0x02, 0x00, 0x00,  // 32
+ 0x00, 0x88, 0x22, 0x04, 0x00, 0x02, 0x00, 0x04,  // 32
  0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  0x00, 0x28, 0x8a, 0x03, 0x00, 0x00, 0x40, 0x00,  // 48
  0x02, 0x00, 0x41, 0xca, 0x9b, 0x06, 0x20, 0x80,
diff --git a/src/tr_languages.cpp b/src/tr_languages.cpp
index 41e85e61..61766220 100644
--- a/src/tr_languages.cpp
+++ b/src/tr_languages.cpp
@@ -45,6 +45,7 @@
 #define OFFSET_GREEK  0x380
 #define OFFSET_CYRILLIC 0x420
 #define OFFSET_DEVANAGARI  0x900
+#define OFFSET_TAMIL  0xb80
 
 
 static const unsigned int replace_cyrillic_latin[] = 
@@ -292,7 +293,7 @@ Translator *SelectTranslator(const char *name)
 			tr->charset_a0 = charsets[19];   // ISCII
 			tr->langopts.length_mods0 = tr->langopts.length_mods;  // don't lengthen vowels in the last syllable
 
-			tr->langopts.stress_rule = 6;      // stress on last heaviest syllable
+			tr->langopts.stress_rule = 6;      // stress on last heaviest syllable, excluding final syllable
 			tr->langopts.stress_flags =  0x10004;   // use 'diminished' for unstressed final syllable
 			tr->langopts.numbers = 0x811;
 			tr->langopts.numbers2 = 0x100;
@@ -314,9 +315,14 @@ Translator *SelectTranslator(const char *name)
 		{
 			static const unsigned char stress_amps_hr[8] = {16,16, 20,20, 20,24, 24,22 };
 			static const short stress_lengths_hr[8] = {180,160, 200,200, 0,0, 220,230};
+			static const short stress_lengths_sr[8] = {160,150, 200,200, 0,0, 250,260};
 
 			tr = new Translator();
-			SetupTranslator(tr,stress_lengths_hr,stress_amps_hr);
+
+			if(name2 == L('s','r'))
+				SetupTranslator(tr,stress_lengths_sr,stress_amps_hr);
+			else
+				SetupTranslator(tr,stress_lengths_hr,stress_amps_hr);
 			tr->charset_a0 = charsets[2];   // ISO-8859-2
 
 			tr->langopts.stress_rule = 0;
@@ -590,6 +596,30 @@ SetLengthMods(tr,3);  // all equal
 		}
 		break;
 
+	case L('t','a'):
+		{
+			static const short stress_lengths_ta[8] = {190, 190,  210, 210,  0, 0,  230, 250};
+			static const unsigned char stress_amps_ta[8] = {17,14, 20,19, 20,24, 24,22 };
+
+			tr = new Translator();
+			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 = 6;      // stress on last heaviest syllable, excluding final syllable
+			tr->langopts.stress_flags =  0x10004;   // use 'diminished' for unstressed final syllable
+			tr->langopts.numbers = 0x811;
+			tr->langopts.numbers2 = 0x100;
+			tr->letter_bits_offset = OFFSET_TAMIL;
+
+			memset(tr->letter_bits,0,sizeof(tr->letter_bits));
+			SetLetterBitsRange(tr,LETTERGP_A,0x05,0x14);   // vowel letters
+			SetLetterBitsRange(tr,LETTERGP_A,0x3e,0x4c);   // vowel signs
+			SetLetterBitsRange(tr,LETTERGP_B,0x3e,0x4d);   // vowel signs, and virama
+			SetLetterBitsRange(tr,LETTERGP_C,0x15,0x39);   // the main consonant range
+			tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1;   // disable check for unpronouncable words
+		}
+		break;
+
 	case L('t','r'):   // Turkish
 		{
 			static const unsigned char stress_amps_tr[8] = {16,16, 20,20, 20,24, 24,22 };
diff --git a/src/translate.cpp b/src/translate.cpp
index 215bb633..27b0586d 100755
--- a/src/translate.cpp
+++ b/src/translate.cpp
@@ -88,8 +88,8 @@ static int embedded_ix;
 static int embedded_read;
 unsigned int embedded_list[N_EMBEDDED_LIST];
 
-// the source text of a single clause
-#define N_TR_SOURCE    350
+// the source text of a single clause (UTF8 bytes)
+#define N_TR_SOURCE    700
 static char source[N_TR_SOURCE+40];     // extra space for embedded command & voice change info at end
 
 int n_replace_phonemes;
@@ -339,12 +339,12 @@ unsigned char *length_mod_tabs[6] = {
  };
 
 
-/* index by 0=. 1=, 2=?, 3=! 4=none */
-static unsigned char punctuation_to_tone[4][5] = {
-	{0,1,2,3,4},
-	{5,6,2,0,4},
-	{0,1,2,3,0},
-	{0,1,2,3,0} };
+/* index by 0=. 1=, 2=?, 3=! 4=none, 5=emphasized */
+static unsigned char punctuation_to_tone[4][6] = {
+	{0,1,2,3,0,4},
+	{5,6,2,0,0,4},
+	{0,1,2,3,0,0},
+	{0,1,2,3,0,0} };
 
 
 void SetLengthMods(Translator *tr, int value)
@@ -369,6 +369,9 @@ int IsAlpha(unsigned int c)
 	if((c >= 0x901) && (c <= 0x957))
 		return(1);    // Devanagari  vowel signs and other signs
 
+	if((c >= 0xb81) && (c <= 0xbe5))
+		return(1);    // Devanagari  vowel signs and other signs
+
 	if((c >= 0x300) && (c <= 0x36f))
 		return(1);   // combining accents
 
diff --git a/src/translate.h b/src/translate.h
index 98688ef3..41405cbb 100755
--- a/src/translate.h
+++ b/src/translate.h
@@ -25,7 +25,7 @@
 
 #define N_WORD_PHONEMES  160          // max phonemes in a word
 #define N_WORD_BYTES     160          // max bytes for the UTF8 characters in a word
-#define N_CLAUSE_WORDS   256          // max words in a clause
+#define N_CLAUSE_WORDS   300          // max words in a clause
 #define N_RULE_GROUP2    120          // max num of two-letter rule chains
 #define N_HASH_DICT     1024
 #define N_CHARSETS        20