[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

dictsource/dict_phonemes

 Dictionary es_dict
 
 @-   a    aI   e    E    eI   eU   i    
-o    oI   u    
+o    O    oI   u    
 
-*    **   b    B    d    D    f    g    
-j    J    J^   k    l    l^   m    n    
-n^   p    Q    r    R    s    t    T    
-tS   v    w    x    
+*    **   :    b    B    d    D    f    
+g    j    J    J^   k    l    l^   m    
+n    n^   p    Q    r    R    s    t    
+T    tS   v    w    x    
 
 
 Dictionary fi_dict
 uai  y    
 
 (u)  :    f    h    j    k    kh   l    
-m    n    N    p    ph   r    s    S#   
-s.   S;   t    th   ts   ts.  ts.  tS;  
-tS;  tsh  v    w    x    z    z.   
+m    n    N    p    ph   r    s    s.   
+S;   t    th   ts   ts.  ts.  tS;  tS;  
+tsh  v    w    x    z    z.   
 
 
 Dictionary jbo_dict

dictsource/es_list

 // letters
 
 _cap	m'aJ^us
-_??	s'imbolo
+_??	s'imbOlo
 _#32	Esp'aTjo
 
 
 // names of symbols
 _.	punto
-_,	koma
-_;	p,untoik'oma
-_:	d,osp'untos
-_!	TE*R'aRaDmi**aTj'on
-_?	TE*R'aRintE*R,oQaTj'on
-_¡	aB@-*'iRaDmi**aTj'on
-_¿	aB@-*'iRintE*R,oQaTj'on
+_,	kOma
+_;	p,untOik'Oma
+_:	d,Osp'untOs
+_!	TE*R'aRaDmi**aTj'On
+_?	TE*R'aRintE*R,OQaTj'On
+_¡	aB@-*'iRaDmi**aTj'On
+_¿	aB@-*'iRintE*R,OQaTj'On
 _=	iQw'al
-_<	mEn'oRke
-_>	maJ^'oRke
-_'	apost@-*'ofo
-_"	kom'iJ^as
-_+	s'iQnodesum'aR
-_$	dolar
-_#	almoaD'il^a
+_<	mEn'ORke
+_>	maJ^'ORke
+_'	apOst@-*'Ofo
+_"	kOm'iJ^as
+_+	s'iQnOdesum'aR
+_$	dOlar
+_#	almOaD'il^a
 _*	astE*'isko
-_-	gJ^on
+_-	gJ^On
 __	suB@-*aJ^'aDo
 _/	ba*Ra
 _\	b'a*RaimbERt'iDa
-_`	aT'Entog@-*'aBe
+_`	aT'EntOg@-*'aBe
 _(	,aB@-*epa**'EntEsis
 _)	Tj,E*Rapa**'EntEsis
-_[	,aB@-*ekoRtS'Ete
-_]	Tj,E*RakoRtS'Ete
+_[	,aB@-*ekORtS'Ete
+_]	Tj,E*RakORtS'Ete
 _{	,aB@-*eJ^'aBe
 _}	Tj,E*RaJ^'aBe
-_«	kom'iJ^as||iTkJ^'ERDas
-_»	kom'iJ^as||dE**'EtSas
+_«	kOm'iJ^as||iTkJ^'ERDas
+_»	kOm'iJ^as||dE**'EtSas
 
 ^	TiRkumfl'Exo
 €	eU*o
-%	porTj'Ento
+%	pOrTj'Ento
 &	ampERs'ant
-@	a*R'oBa
+@	a*R'OBa
 /	ba*Ra
-©	kopi*R'aIt
+©	kOpi*R'aIt
 £	liB@-*as
 ¶	p'a*Rafo
-§	sEkTj'on
+§	sEkTj'On
 
 
 // numbers
 _0	TE*o
 _1	'uno
-_2	d'os
+_2	d'Os
 _3	t@-*'es
 _4	kw'at@-*o
 _5	T'inko
 _6	s'eIs
 _7	sj'Ete
-_8	'otSo
+_8	'OtSo
 _9	nw'Eve
 _1X	dj'ET
-_11	'onTe
-_12	d'oTe
+_11	'OnTe
+_12	d'OTe
 _13	t@-*'ETe
-_14	kat'oRTe
+_14	kat'ORTe
 _15	k'inTe
 _20	v'eInte
 _2X	v'eInt
 _5X	Tinkw'Enta
 _6X	sEs'Enta
 _7X	sEt'Enta
-_8X	otS'Enta
-_9X	nov'Enta
-_0C	T'ientos
+_8X	OtS'Enta
+_9X	nOv'Enta
+_0C	T'ientOs
 _1C0	T'ien         // exactly one hundred
 _1C	T'iento
-_5C	kinj'Entos
-_7C	s,EtETj'Entos
-_9C	n,ovETj'Entos
+_5C	kinj'EntOs
+_7C	s,EtETj'EntOs
+_9C	n,OvETj'EntOs
 _0M1	m'il
 _1M1	m'il          // no '1' before thousand
-_0M2	mil^'ones
-_1M2	'unmil^'on
-_0M4	_bil^'ones
-_1M4	'unbil^'on
+_0M2	mil^'Ones
+_1M2	'unmil^'On
+_0M4	_bil^'Ones
+_1M4	'unbil^'On
 _0and	i
-_dpt	koma
-//_roman	Rom'ano
+_dpt	kOma
+//_roman	ROm'ano
 
 vi	vi	// not a Roman number
 

dictsource/es_rules

 
 
 .group j
-       _) j (_       x'ota
+       _) j (_       x'Ota
           j          x
 
 
 
 .group o
        _) o (_       o
-          o          o
+          o          O
+          o (_       o
           oy (K      oI
           oy (_      'oI
 
 .group p
        _) p (_       pe
           p          p
+          p (t       p:
 
 
 .group q
 
 
 .group w
-       _) w (_       ,uBe||d'oBle
+       _) w (_       ,uBe||d'OBle
           w          w
 
 
 .group x
        _) x (_       Ekis
           x          ks
-       A) x (A       Qs
+//       A) x (A       Qs  // ??
           x (_       =Qs
 
 
           á          ''a
           é          ''E
           í          ''i
-          ó          ''o
+          ó          ''O
           ú          ''u
           ü          u
           ñ          n^
         _)ñ (_       En^e
 
-          $          d'olar
+          $          d'Olar
 
           +          m'as
       _)  ++ (_      masm'as
  \+\+\+)  +          // ignore + after the first 3
 
-          #          almoaD'il^a
+          #          almOaD'il^a
       \#) #
 
-     __)  - (_D      m'Enos
+     __)  - (_D      m'EnOs
      A_)  - (_D      _
      C_)  - (_D      _
       --) -
           . (.       _:
       \.) .
 
-          , (D       koma
+          , (D       kOma

dictsource/fr_list

 
 // 2006-11-18  Gilles Casse  <[email protected]>
 //
-// Updated 2008-02-08 Michel Such  <[email protected]>
+// Updated 2008-02-10 Michel Such  <[email protected]>
 //
 //	* Numbers, a few abbreviations and exceptions.
 //
 adsl		$abbrev
 alt $pause
 amd		$abbrev
+anpe             $abbrev
 api		$abbrev
 avh		$abbrev
 cci              $abbrev
 mr		m@sjY
 nti             $abbrev
 nvda             $abbrev
+ong             $abbrev
 pci             $abbrev
 pda             $abbrev
 p.s		postskRiptOm
 accident      aksidA~
 adéquat       adekuat2
 adéquate      adekuat
+airbus	      Erbys
+anus     anys
+any     Eni
 ardent        aRdA~
+bissus    bisys
+bonus     bonys
 bus	      bys
+bus	      by  $verb
+campus   kA~pys
+casus    kazys
 choeur        k@r
+cocus     coky
 discident     disidA~
+eye	      aj
+eyes	      ajz
 fils	      fis
 imprudent     E~pRydA~
 juin	      ZyE~
+malus      malys
+many     mEni
 mars	      maRs
+minus      minys
 occident      OksidA~
 orient        ORjA~
 pays	      pEi
+phallus    falys
 prudent       pRydA~
 secret        sYkRE
 souris       suri
-
+sus       sys
+sus       sy $verb
+to       tu
+vénus       venys
 
 
 mens	      mA~z2
 
 // words from other languages
 about		@bawt
+acer             asEr
 amazon		amaz'On
 (audible manager) od'ibl||manadZ'@r
 be		bi
 klaxon          klaksOn
 leader          li:d@R
 let 		lEt
-link 		link
 liszt 		list
 lite 		lajt
 (mac os x)	makoEsiks
 schubert	SubER
 screamer		skri:m@r
 (search & destroy) s@:rS_||end_||dEsstrOj
-shrink		Sr'ink
 skype 		sk'ajp@-
 spybot		spajbOt
 sun		s@n

dictsource/fr_rules

 
 // 2006-11-18  Gilles Casse  <[email protected]>
 //
-// Updated: 2008-02-08 Michel Such  <[email protected]>
+// Updated: 2008-02-10 Michel Such  <[email protected]>
 //
 //	* The rules are based on Cicero TTS.
 //
 		em (me      a		// femme patiemment
 		em (b       A~		// décembre
 		em (m       A~		// emmancher
+		em (manu       e		// emmanuel
 		em (p       A~
 		en (_       A~n2		// en
 		en (hA      A~n		// enharmonique
 _)		h	(arf	_!		// harfang
 _)		h	(argn	_!		// hargne
 _)		h	(ari	_!		// haricot haridelle
+_)		h	(ark	_!		// harka
 _)		h	(arn	_!		// harnais
 _)		h	(arp	_!		// harpe
 _)		h	(as	_!		// hasard, hase
 _)		h	(âs	_!		// hâte
 _)		h	(au	_!		// haut
 _)		h	(av	_!		// havre
+
 _)		h	(eau	_!		// heaume
 _)		h	(él	_!		// hélas, héler
 _)		h	(èl	_!		// hèlement
 _)		h	(êtr	_!		// hêtre, hêtraie
 _)		h	(eula	_!		// heulandite
 _)		h	(eurt	_!		// heurter
+
 _)		h	(iat	_!		// hiatal, hiatus
 _)		h	(ibo	_!		// hibou
 _)		h	(ic	_!		// hic
 _)		h	(i-han	_!		// hi-han
 _)		h	(iss	_!		// hisser
 _)		h	(it	_!		// hit, hit-parad
+
 _)		h	(ob	_!		// hobby, hobereau
 _)		h	(och	_!		// hocher
 _)		h	(ock	_!		// hockey
 _)		h	(ou	_!		// houle, housse
 
 _)		h	(u	_!
+_)		h	(uil			// huile
 _)		h	(uiss			// huissier
 _)		h	(uî			// huître
 _)		h	(umain			// humain
 		ing (_      iN		// parking
 		in (h       in		// inhumain
 		inct (_     E~		// distinct
+		ink (_     ink		// link, pink
 		in (n       i		// innombrable
 		in (C       E~		// vingt
 		    //		in (_       E~		// vin
 		o (che_       O		// poche, cloche
 		oe (ll      wa		// moelleux
 	_)	oe (C       Y		// oesophage
+	_ph)	oe (t       Y		// phoetus
 		oeu       @		// soeur coeur
 		oeufs       @		// oeufs boeufs
 		o (eC       o		// coefficient
 	xpan) 	s (A        s		// expansion
 
 
-	ibu) 	s (_        s
+	_u) 	s (_        s
+	Cu) 	s (_
+	Abu) 	s (_        s
+	_Abu) 	s (_
 	mbu) 	s (_        s
-
-	cu) 	s (_        s		// focus phallus cumulus minus
-	iu) 	s (_        s
+	Acu) 	s (_        s
+	écu) 	s (_
+	odu) 	s (_        s
+	gu) 	s (_        s
+	igu) 	s (_               // aigus
+	iu) 	s (_        s      // radius
+	ïu) 	s (_        s      // laïus
+	Culu) 	s (_        s   // cumulus
+	plu) 	s (_        z2     // plus, surplus
 	mu) 	s (_        s
-	nu) 	s (_        s
-	pu) 	s (_        s
-	llu) 	s (_        s
-	ulu) 	s (_        s
+	_mu) 	s (_
+	ému) 	s (_
+	opu) 	s (_        s // opus, octopus
+	Aru) 	s (_        s // virus
+	aru) 	s (_         // disparus
+	psu) 	s (_        s // lapsus
 
 	atu) 	s (_        s		// stratus cactus motus
-	ctu) 	s (_        s
+	ctu) 	s (_        s     // cactus
+	etu) 	s (_        s
+	aitu) 	s (_
 	itu) 	s (_        s
-	otu) 	s (_        s
+	ltu) 	s (_        s     // altus
+	otu) 	s (_        s     // motus
 
 	_tou)	s (_C
 	_tou) 	s (_        s
 		u           y	// cruel nuage brut
 		ui          w^i
 
+	_b)	usi       iz  // business, businessman
 	c)	ueill       Wj  // cueillir
 	c) 	ueil (_     Wj  // recueil
+	_c) 	ul (_     y
 	g)	ueill       Wj	// orgueilleux
 	g) 	ueil (_     Wj	// orgueil
 
 		ym (p       E~	// tympan
 		y (nn       i
 		y (nA       i
+		ys (_       is
 		yx (_       iks
 
 		yn          E~	// laryngite

dictsource/zh_rules

 .group y
        y	j //before a o e i
        y (u	//NULL before u
+       y (uDn   j
        yo (DngK	yu
        y (K     i    // foreign words
 

phsource/compile_report

       is  32  124
       vi  42  137
      zhy  40  132
-      zh  64  148
+      zh  63  147
       sw  14  108
       th  50  142
       id  14  120
                 [s#] pt_pt
 ufric/sh_pzd    [S;] base
                 [S;] ro
-                [S;] zh
 ufric/sh_pzd_   [S;] base
                 [Z;] base
-ufric/sh_rfx    [s.] base
-                [s.] zh
-ufric/sh_sr     [S#] zh
-ufric/sh_sr.wav  [S] sr
+ufric/sh_pzd2   [S] sr
                 [Z] sr
                 [dZ] sr
+                [S;] zh
+ufric/sh_rfx    [s.] base
+                [s.] zh
 ufric/s_pzd     [s;] base
                 [z;] base
                 [s;] ro
                 [ch] th
 ustop/tsh_pzd_unasp  [tS;] zh
                 [c] th
-ustop/tsh_sr.wav  [tS] sr
+ustop/tsh_sr    [tS] sr
 ustop/ts_pzd    [tS;] base
                 [ts;] base
                 [ch] hi
                 [t] zh
 ustop/ts_rfx    [ts.] zh
 ustop/ts_rfx_unasp  [ts.] zh
-ustop/ts_sr.wav  [tS;] sr
+ustop/ts_sr     [tS;] sr
                 [dZ;] sr
 ustop/ts_unasp  [ts] zh
 ustop/t_unasp   [t] th
                 [u@] vi
                 [ua] th
                 [u:a] th
+vdiph2/y@       [y@] zh
 vdiph2/y#@      [Y@] af
-                [y@] zh
 vdiph2/yu       [yU] nl
                 [yu] zh
 vdiph2/y-y#     [yY] fi
                 [O@] en_wi
                 [O] de
                 [o] eo
-                [o] es
+                [O] es
                 [o] el
                 [O] sv
                 [O] no

phsource/ph_serbian

-// ufric/x_sr
-// ufric/ch_sr
-// ufric/sh_sr
-// ustop/t_sr
-// ustop/ts_sr
-// ustop/tsh_sr
-// r3/r_sr
-// m/m_sr
-// n/n_sr
-// n^/nj_sr
 
 // This file inherits phonemes from ph_croatian
 // Only phonemes which differ from the inherited phonemes need
   vcd pla afr sibilant
   vowelin f1=2  f2=2300 200 400  f3=100 80
   vowelout f1=2  f2=2300 250 300  f3=100 80 brk
-  formants dzh/dzh+ufric/sh_sr.wav
-  before _ dzh/dzh+ufric/sh_sr.wav
+  formants dzh/dzh+ufric/sh_pzd2
+  before _ dzh/dzh+ufric/sh_pzd2
   after _ NULL
-  after @ dzh/dzh+ufric/sh_sr.wav
+  after @ dzh/dzh+ufric/sh_pzd2
   lengthmod 5
   switchvoicing tS
 endphoneme
 phoneme dZ;
   vcd pla afr sibilant palatal
   vowelin f1=2  f2=2700 400 600  f3=300 80 rate len=70
-  formants dzh/dzh+ustop/ts_sr.wav
-  before _ dzh/dzh+ustop/ts_sr.wav
+  formants dzh/dzh+ustop/ts_sr
+  before _ dzh/dzh+ustop/ts_sr
   after _ NULL
-  after @ dzh/dzh+ustop/ts_sr.wav
+  after @ dzh/dzh+ustop/ts_sr
   lengthmod 5
   switchvoicing tS;
 endphoneme
   vls pla afr sibilant
   vowelin f1=0  f2=2300 200 400  f3=-100 80
   lengthmod 2
-  wave ustop/tsh_sr.wav%50
-  before _ ustop/tsh_sr.wav%50
+  wave ustop/tsh_sr%50
+  before _ ustop/tsh_sr%50
   switchvoicing dZ
 endphoneme
 
   vls pal afr sibilant palatal
   vowelin f1=0  f2=2700 400 600  f3=300 80 rate len=70
   lengthmod 2
-  wave ustop/ts_sr.wav
-  before _ ustop/ts_sr.wav
+  wave ustop/ts_sr
+  before _ ustop/ts_sr
   switchvoicing dZ;
 endphoneme
 
   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%60
-  before _ voc/zh+ufric/sh_sr.wav%50
+  formants voc/zh+ufric/sh_pzd2%60
+  before _ voc/zh+ufric/sh_pzd2%50
   lengthmod 6
   switchvoicing S
 endphoneme
   vowelout f1=0  f2=2100 100 300  f3=-100 80
   length 100
   lengthmod 3
-  wave ufric/sh_sr.wav%70
-  before _ ufric/sh_sr.wav%60
+  wave ufric/sh_pzd2%70
+  before _ ufric/sh_pzd2%60
   switchvoicing Z
 endphoneme
 

phsource/ph_spanish

   formants vowel/e_mid
 endphoneme
 
-phoneme o
+phoneme O
   vowel starttype (o) endtype (o)
   length 170
   formants vowel/oo

phsource/ph_zh

   vls pal afr sibilant palatal
   vowelin f1=0  f2=2700 400 600  f3=300 80
   lengthmod 2
-  wave ustop/tsh_pzd
+  wave ustop/tsh_pzd%80
 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 S#
-  vls pla frc sibilant
-  vowelin f1=0  f2=2700 400 600  f3=300 80
-  lengthmod 2
-  wave ufric/sh_sr%20
+  wave ufric/sh_pzd2%20
 endphoneme
 
 phoneme s.
   vowelout f1=0  f2=1800 -100 300  f3=-300 80
   length 100
   lengthmod 3
-  wave ufric/sh_rfx%60   // quieter
+  wave ufric/sh_rfx%50   // quieter
   before _ ufric/sh_rfx%50
 endphoneme
 
 phoneme y@
   vowel  starttype (i)  endtype (@)
   length 250
-  formants vdiph2/y#@
+  formants vdiph2/y@
 endphoneme

phsource/phonemes

 phoneme **    //   TEST flap
   liquid alv
   vowelout f1=3 f2=1600 -300 300  f3=-300 80 rms=35 len=15
-  vowelin f1=2  f2=1600 -300 300  f3=-300 80
-  formants r3/@tap+r3/rx%50
+  vowelin f1=2  f2=1600 -300 300  f3=-100 80 len=20
+  formants r3/@tap
   vowelin f1=3  f2=1600 -300 300  f3=-300 80
   before (i) r3/@tap_i+r3/rx
   before (e) r3/@tap_i+r3/rx

src/intonation.cpp

 };
 
 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
 		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)
 					}
 				}
 
-				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

 	{"amp", '&'},
 	{"quot", '"'},
 	{"nbsp", ' '},
+	{"apos", '\''},
 	{NULL,-1}};
 
 
 	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;
 	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))
 		{
 
 		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));
-					else
-						sscanf(&buf2[1],"%d",&c1);
-				}
-				else
-				{
-					if((j = LookupMnem(xml_char_mnemonics,buf2)) != -1)
+					if(buf2[0] == '#')
 					{
-						c1 = j;
+						// character code number
+						if(buf2[1] == 'x')
+							found = sscanf(&buf2[2],"%x",(unsigned int *)(&c1));
+						else
+							found = sscanf(&buf2[1],"%d",&c1);
 					}
 					else
 					{
-						c1 = '&';
+						if((found = LookupMnem(xml_char_mnemonics,buf2)) != -1)
+						{
+							c1 = found;
+							if(c2 == 0)
+								c2 = ' ';
+						}
 					}
 				}
+				else
+				{
+					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

 			else
 				p->prepause = 60;
 
+			if(prev->type == phSTOP)
+				p->prepause = 60;
+
 			if((langopts.word_gap & 0x10) && (p->newword))
 				p->prepause = 60;
 

src/synthdata.cpp

 #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;