Belarusian: improvements (#1417)

ChangeLog.md

 
 updated languages:
 *  ba (Bashkir) -- Andiv06
-*  be (Belarusian) -- Andiv06
+*  be (Belarusian) -- Andiv06, bespsm, Alaksiej Stankievič
 *  cmn (Mandarin) -- Cameron Wong
 *  en (English) -- Bill Dengler
 *  es (Spanish) -- Sukil Etxenike

dictsource/be_list

 // Spelling-to-phoneme words for Belarusian
 
 // Letter names
-а       a
-б       bE
-в       vE
-г       QE
-д       dE
-е       jE
-ё       jO
-ж       z.E
-з       zE
-і       i
-й       i||n;EskladOvajE
-к       ka
-л       El
-м       Em
-н       En
-о       O
-п       pE
-р       Er
-с       Es
-т       tE
-у       u
-ў       u||n;EskladOvajE
-ф       Ef
-х       xa
-ц       tsE
-ч       ts.E
-ш       s.a
-ы       i"
-ь       m;ak;:i||znak
-э       E
-ю       ju
-я       ja
-'       apOstraf
+а       a $u
+б       bE $u
+в       vE $u
+г       QE $u
+д       dE $u
+е       jE $u
+ё       jO $u
+ж       z.E $u
+з       z $u
+і       i $u
+й       i||n;Esklad'OvajE
+к       ka $u
+л       El  $u
+м       Em $u
+н       En $u
+о       O   $u
+п       pE  $u
+р       Er  $u
+с       Es  $u
+т       tE  $u
+у       u $u
+ў       u||n;Esklad'OvajE
+ф       Ef $u
+х       xa $u
+ц       tsE $u
+ч       ts.E $u
+ш       s.a $u
+ы       i" $u
+ь       m;'ak;:i||znak
+э       E $u
+ю       ju $u
+я       ja $u
 
 // Numbers
-_0     nul;
-_1     ad;in
-_2     dva
-_3     tri"
-_4     ts.ati"ri"
-_5     p;at;
-_6     s.Es;t;
-_7     s;Em
-_8     vOs;E2m
-_9     d;Ev;at;
-_10    d;Es;at;
-_11    ad;inats:at;
-_12    dvanats:at;
-_13    tri"nats:at;
-_14    ts.ati"rnats:at;
-_15    p;atnats:at;
-_16    s.asnats:at;
-_17    s;amnats:at;
-_18    vas;amnats:at;
-_19    d;Ev;atnats:at;
-_2X    dvats:at;
-_3X    tri"ts:at;
-_4X    sorak
-_5X    p;adz;:Es;at
-_6X    s.Ezdz;:Es;jat
-_7X    s;Emdz;Es;jat
-_8X    vos;Emdz;Es;at
-_9X    d;Ev;anosta
-_0C    stO
-_2C    d;v;Es;t;E
-_3C    tri"sta
-_4C    ts.ati"ri"sta
-_5C    p;at;sot
-_6C    s.Es;t;sot
-_7C    s;jEmsot
-_8C    vos;jEmsot
-_9C    d;Ev;at;sot
-_1MA1  ti"s;ats.a
-_0MA1  ti"s;ats.i"
-_0M1   ti"s;ats.
+_0     n'ul;
+_1     adz;'in
+_1f    adna
+_2     dv'a
+_2f    dz;v;'E
+_3     tr'i"
+_4     ts.at'i"ri"
+_5     p;'ats;
+_6     s.'Es;ts;
+_7     s;'Em
+_8     v'Os;Em
+_9     dz;'Ev;ats;
+_10    dz;'Es;ats;
+_11    adz;'inats:ats;
+_12    dvan'ats:ats;
+_13    tri"n'ats:ats;
+_14    ts.at'i"rnats:ats;
+_15    p;atn'ats:ats;
+_16    s.asn'ats:ats;
+_17    s;amn'ats:ats;
+_18    vas;amn'ats:ats;
+_19    dz;Ev;atn'ats:ats;
+_2X    dv'ats:ats;||
+_3X    tr'i"ts:ats;||
+_4X    s'Orak||
+_5X    p;adz;:Es;'at||
+_6X    s.Ezdz;:Es;'at||
+_7X    s;'Emdz;Es;at||
+_8X    v'Os;Emdz;Es;at||
+_9X    dz;Ev;an'Osta||
+_0C    st'O
+_2C    dz;v;'Es;ts;E
+_3C    tr'i"sta
+_4C    ts.at'i"ri"sta
+_5C    p;ats;s'Ot
+_6C    s.Es;ts;s'Ot
+_7C    s;Ems'Ot
+_8C    v,Os;Ems'Ot
+_9C    dz;Ev;ats;s'Ot
+_0MB1  t'i"s;ats.a
+_0MA1  t'i"s;ats.i"
+_0M1   t'i"s;ats.      // NOTE: in 10k > appends `_!`

dictsource/be_rules

 // This file is UTF8 encoded
 // Translation rules for Belarusian
 
+// NOTE: LETTERGP_A doesn't work, L01 as replacement
+//letter which can carry vowel phoneme
+.L01 а е ё і о у ы э ю я
+
+// This is more economic way to express assimilation by sibilation
+.replace
+       сш     шш
+       сч     шч
+       цч     чч
+       шс     сс
+       шц     сц
+       чц     цц
+       зж     жж
+       здж    ждж
+       жз     зз
+       ждз    здз
+
 .group а
        а       a
 
 
 .group д
        д       d
+       дз      dz
+       дж      dz.
 
 .group е
        е       ;E    // ʲɛ
+    _) е       jE
+    ') е       jE
+    ь) е       jE
+  L01) е       jE
 
 .group ё
-       ё       ;O    // ʲɔ
+       ё       ;'O    // ʲ'ɔ
+    _) ё       j'O
+    ') ё       j'O
+    ь) ё       j'O
+  L01) ё       j'O
 
 .group ж
        ж       z.    // ʐ
 
 .group і
        і       i
+    C) і       ;i
 
 .group й
        й       j
        н       n
 
 .group о
-       о       o
+       о       'O
 
 .group п
        п       p
        х       x
 
 .group ц
-       ц       ts    // t̻͡s
+       ц       ts    // t͡s
+       цц      ts:
 
 .group ч
        ч       ts.   // ʈ͡ʂ
+       чч      ts.:
 
 .group ш
        ш       s.    // ʂ
+       шш      s.:
 
 .group ы
-       ы       i"    // (ɨ
+       ы       i"    // ɨ
 
 .group ь
        ь       ;     // ʲ
 
 .group ю
        ю       ;u    // ʲu
+    _) ю       ju
+    ') ю       ju
+    ь) ю       ju
+  L01) ю       ju
 
 .group я
        я       ;a    // ʲa
+    _) я       ja
+    ') я       ja
+    ь) я       ja
+  L01) я       ja
 
 .group '
-       '       j
+       '

espeak-ng-data/lang/zle/be

 name Belarusian
 language be
-replace 03 a a#
-dict_min  20000
+dict_min  2000
 speed 95

phsource/ph_belarusian

 //  Belarusian
 //====================================================
 
-
 phoneme a
-  import_phoneme ru/a
+  vwl   starttype #a  endtype #a
+  ipa a
+  length 180
+  FMT(vowel/a)
 endphoneme
 
-phoneme O
-  vwl starttype #o endtype #o
-  ipa ɔ
-  length 100
-  FMT(vwl_ru/8)
+phoneme i"
+  vwl starttype #i endtype #i
+  ipa ɨ
+  length 180
+  FMT(vowel/ii#_2)
 endphoneme
 
-phoneme i"
-  import_phoneme pl/y
+phoneme Q
+  import_phoneme base1/Q"
+  ipa ɣ
 endphoneme
 
 phoneme ts
-  import_phoneme consonants/ts
-  ipa t̻͡s
+  CALL consonants/ts
+  voicingswitch dz
+  ipa t͡s
 endphoneme
 
 phoneme ts;
-  import_phoneme pl/ts;
-  ipa t̻͡s̪ʲ
+  CALL base1/tS;
+  voicingswitch dz;
+  ipa t͡sʲ
 endphoneme
 
 phoneme ts.
-  import_phoneme sr/tS
+  vls pla afr sib
   ipa ʈ͡ʂ
+  lengthmod 2
+  voicingswitch dz.
+  Vowelin f1=0  f2=2300 200 400  f3=-100 80
+  WAV(ustop/tsh_sr, 50)
 endphoneme
 
 phoneme dz
-  import_phoneme consonants/dz
-  ipa d̻͡z̪
+  CALL consonants/dz
+  voicingswitch ts
+  ipa d͡z
 endphoneme
 
 phoneme dz;
   import_phoneme pl/dz;
-  ipa d̻͡z̪ʲ
+  CALL base1/dZ;
+  voicingswitch ts;
+  ipa d͡zʲ
 endphoneme
 
 phoneme dz.
-  import_phoneme sr/dZ
+  vcd pla afr sib
   ipa ɖ͡ʐ
+  lengthmod 5
+  voicingswitch ts.
+  Vowelin f1=2  f2=2300 200 400  f3=100 80
+  Vowelout f1=2  f2=2300 250 300  f3=100 80 brk
+endphoneme
+
+phoneme r
+  liquid trl
+  lengthmod 6
+  ipa r
+  length 80
+  FMT(r3/r_trill2) addWav(r3/r_trill2.wav, 65)
+endphoneme
+
+phoneme ;
+  liquid pzd
+  lengthmod 0
+  ipa ʲ
 endphoneme

phsource/phonemes

 
 // *** Tables extending base2 ***
 
+phonemetable be base2
+include ph_belarusian
+
 phonemetable chr base2
 include ph_cherokee
 
 phonemetable qdb en
 include ph_langbelta
 
-phonemetable be ru
-include ph_belarusian
-
 phonemetable ms id

src/libespeak-ng/tr_languages.c

 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_6;
 		SetArabicLetters(tr);
 		break;
+	case L('b', 'e'): // Belarusian
+	{
+		static const unsigned char stress_amps_be[8] = { 12, 10, 8, 8, 0, 0, 16, 17 };
+		static const short stress_lengths_be[8] = { 160, 140, 200, 140, 0, 0, 240, 160 };
+		static wchar_t vowels_be[] = { // offset by 0x420 -- а е ё о у ы э ю я і
+			0x10, 0x15, 0x31, 0x1e, 0x23, 0x2b, 0x2d, 0x2e, 0x2f, 0x36, 0
+		};
+		static const unsigned char consonants_be[] = { // б в г д ж з й к л м н п р с т ф х ц ч ш ў
+			0x11, 0x12, 0x13, 0x14, 0x16, 0x17, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1f, 0x20, 0x21, 0x22, 0x24, 0x25, 0x26, 0x27, 0x28, 0x3e, 0
+		};
+
+		tr->langopts.stress_flags = S_NO_AUTO_2 | S_NO_DIM; // don't use secondary stress
+		tr->letter_bits_offset = OFFSET_CYRILLIC;
+		tr->transpose_min = 0x430;  // convert cyrillic from unicode into range 0x01 to 0x2f
+		tr->transpose_max = 0x45e;
+		memset(tr->letter_bits, 0, sizeof(tr->letter_bits));
+		SetLetterBits(tr, LETTERGP_A, (char *)vowels_be);
+		SetLetterBits(tr, LETTERGP_C, (char *)consonants_be);
+		SetLetterBits(tr, LETTERGP_VOWEL2, (char *)vowels_be);
+
+		SetupTranslator(tr, stress_lengths_be, stress_amps_be);
+		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_5;
+		tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1; // disable check for unpronouncable words
+		tr->langopts.stress_rule = STRESSPOSN_1L;
+		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_OMIT_1_HUNDRED;
+		tr->langopts.numbers2 = NUM2_THOUSANDPLEX_VAR_THOUSANDS | NUM2_THOUSANDS_VAR1; // variant numbers before thousands
+	}
+		break;
 	case L('b', 'g'): // Bulgarian
 	{
 		SetCyrillicLetters(tr);

tests/language-numbers-cardinal.test

 
 # European Style (hundred, thousand, -illion/-illiard)
 
+echo "... be (Belarusian)" # short scale
+
+test_phon be "n'ul; adz;'in dv'a tr'i\" ts.at'i\"ri\" p;'ats; s.'Es;ts; s;'Em v'Os;Em dz;'Ev;ats;" "0 1 2 3 4 5 6 7 8 9" "units"
+test_phon be "dz;'Es;ats; adz;'inats:ats; dvan'ats:ats; tri\"n'ats:ats; ts.at'i\"rnats:ats; p;atn'ats:ats; s.asn'ats:ats; s;amn'ats:ats; vas;amn'ats:ats; dz;Ev;atn'ats:ats;" "10 11 12 13 14 15 16 17 18 19" "teens"
+test_phon be "dv'ats:ats; adz;'in tr'i\"ts:ats; s'Orak p;adz;:Es;'at s.Ezdz;:Es;'at s;'Emdz;Es;at v'Os;Emdz;Es;at dz;Ev;an'Osta" "21 30 40 50 60 70 80 90" "tens"
+test_phon be "st'O adz;'inats:ats; dz;v;'Es;ts;E dv'ats:ats; dv'a tr'i\"sta ts.at'i\"ri\"sta p;ats;s'Ot s.Es;ts;s'Ot s;Ems'Ot v,Os;Ems'Ot dz;Ev;ats;s'Ot" "111 222 300 400 500 600 700 800 900" "hundreds"
+test_phon be "adna t'i\"s;ats.a dz;v;'Es;ts;E tr'i\"ts:ats; ts.at'i\"ri\" dz;v;'E t'i\"s;ats.i\" tr'i\"sta s'Orak tr'i\" t'i\"s;ats.i\" ts.at'i\"ri\"sta ts.at'i\"ri\" t'i\"s;ats.i\" p;'ats; t'i\"s;ats. dz;'Es;ats; t'i\"s;ats. dz;'Es;ats; t'i\"s;ats._! adz;'in" "1234 2340 3400 4000 5000 10000 10001" "thousands"
+
 echo "... en (English)" # short scale
 
 test_phon en "z'i@roU w'0n t'u: Tr'i: f'o@ f'aIv s'Iks s'Ev@n 'eIt n'aIn" "0 1 2 3 4 5 6 7 8 9" "units"

tests/language-phonemes.test

 test_phwav as 05d4cca91fc3447ae8b6acd7892790dd364e8e23 "ma na Na pa ta ka p#a t#a k#a ba da ga b#a d#a g#a sa xa Xa ha tSa tS#a za wa ra ja la _:_ ma mi mu me m& mo mO mV ma~ mi~ mu~ me~ m&~ mo~ mO~ moj mo~j mow mo~w mew maV m@"
 test_phwav az 7bdb78c37433a47fb3b15808071b1a916202400c "ma na pa ba ta da tSa dZa tsa dza ca Ja Ca ka ga fa va sa za Sa Za xa Qa ha la ja wa Ra *a _:_ mi me m& my mW mu mo ma m@"
 test_phwav ba cf21e5edc1227d3483f49305fbb39e201e5a7f97 "ma na pa ba ta da tSa dZa tsa dza ca Ja Ca ka ga fa va sa za Sa Za xa Qa ha la ja wa Ra *a _:_ mi me m& my mW mu mo ma m@"
-test_phwav be 88cd154560ce202e65aa281ac7b62454806c7c6c "a O i\" tse ts;e ts.i dzi dz;o dz.u"
+test_phwav be 46e35280b05300c9b7ed2a6dd0434c299689b673 "a i\" Q;E Qu tsO ts;E ts:E ts.u ts.O ts.;E ts.:a dzO dz;a dz.O dz.u"
 test_phwav bg 3020acb23ac0f93d5479de3b305b71fc07b5e738 "ma na n^a Na m;a pa ba ta da ka ga p;a b;a t;a d;a ca Ja tsa dza tSa dZa ts;a dz;a fa va sa za Sa Za xa Qa f;a v;a s;a z;a x;a Ra R;a wa ja l/2a la l^a _:_ mi me m@ ma mo mu"
 test_phwav bn 7fc6c8c6b347677885c6e973ff4dc6b2b6fa6d59 "ma na Na pa ta t.a tSa ka p#a t#a t.#a tS#a k#a ba da d.a dZa ga b#a d#a d.#a g#a fa Ba sa za Sa Za ha Ha wa la ja ra *a Ra _:_ ma mE mO me mi mi mu m& mV"
 test_phwav bpy 7fc6c8c6b347677885c6e973ff4dc6b2b6fa6d59 "ma na Na pa ta t.a tSa ka p#a t#a t.#a tS#a k#a ba da d.a dZa ga b#a d#a d.#a g#a fa Ba sa za Sa Za ha Ha wa la ja ra *a Ra _:_ ma mE mO me mi mi mu m& mV"

tests/language-pronunciation.test

 test_phon az "byts'yn insanL'aR l&jag'&t v& hygugLa*@n'a J'W*& az'ad b&*ab'&R do:uLuRL'aR
 ona*'@n Syu*aLR'@ v& vidZdanLa*'@ v'aR v& b'iRbiRl&*'in& mynasib'&td& gaRdaSL'@x Runhund'a davRanmaL@d@RL'aR" "Bütün insanlar ləyaqət və hüquqlarına görə azad bərabər doğulurlar. Onarın şüuralrı və vicdanları var və bir-birlərinə münasibətdə qardaşlıq runhunda davranmalıdırlar." "Latn"
 test_phon ba "m'in h'in 'uL b'eD h'eD 'uLAr b'VL t'eQe" "мин һин ул беҙ һеҙ улар был теге." "Cyrl"
+test_phon be "Qr'us.a tsv;'ila ap'Os.n;i Q'Od
+'us;E Q'al;ini\" j'ajE
+'us;E v;'al;ik;ija ras'Ox;i
+d'a ap'Os.n;aQa pr'uts;ika
+b'i\"l;i ws'i\"pani\" b'ujni\"m b;'Elaruz.'Ovi\"m tsv;'Etam" "Груша цвіла апошні год. Усе галіны яе, усе вялікія расохі, да апошняга пруціка, былі ўсыпаны буйным бела-ружовым цветам." "Cyrl"
 test_phon bg "'ax tS'udna b@Lg'arska z'em;o
 pol'uSvaj ts@ft;'aSti: Z'ita" "Ах чудна българска земьо, полюшвай цъфтящи жита." "Cyrl"
 test_phon bn "m'alOS,ij 'Obojd#,O Sr'omik,Oder b'i*udd#,e tS'OlOm,an_:_: m'egat#r'i_:_: 'ob#idZ,ane Se d'eSe S'Oto S'Oto 'ob#ib,aSik,e 'at.Ok k'O*etS#,e Se d'eSer 'imigr,eSOn p'uliS