Merge remote-tracking branch 'jaacoppi/maintainability'

docs/voices.md

   - [dictionary](#dictionary)
   - [dictrules](#dictrules)
   - [replace](#replace)
+  - [stressRule](#stressrule)
   - [stressLength](#stresslength)
   - [stressAdd](#stressadd)
   - [stressAmp](#stressamp)
   - [intonation](#intonation)
   - [dictmin](#dictmin)
-
+  - [letterVowel](#letterVowel)
 ----------
 
 A Voice file specifies a language (and possibly a language variant or
 The phoneme mnemonics can be defined for each language, but some are
 listed in [Phonemes](phonemes.md).
 
+### stressRule
+
+	stressRule <4 integer values>
+
+Four integer parameters. These correspond to:
+
+1. langopts->stress_rule (values in translate.h)
+2. langopts->stress_flags
+3. langopts->unstressed_wd1 (stress for $u word of 1 syllable)
+4. langopts->unstressed_wd2 (stress for $u word of >1 syllable) 
+
+All parameters are not required. For example, "stressRule 2 1" is valid.
+
 ### stressLength
 
 	stressLength <8 integer values>
 installed. If the size of the compiled dictionary data for the language
 (the file `espeak-ng-data/*_dict`) is less than this size then a
 warning is given.
+
+### letterVowel
+
+	letterVowel <letter>
+
+Used for some languages to handle a certain letter as a vowel instead of consonant.

espeak-ng-data/lang/aav/vi

 
 words 1
 pitch 80 118
+stressRule 0

espeak-ng-data/lang/art/eo

 language eo
 
 apostrophe 2
+stressRule 2

espeak-ng-data/lang/art/ia

 name Interlingua
 language ia
+stressRule 2

espeak-ng-data/lang/art/jbo

 language jbo
 
 speed 80   // speed adjustment, percentage
+letterVowel y
+stressRule 2

espeak-ng-data/lang/azc/nci

 language nci
 
 intonation 3
-stressrule 2
+stressRule 2
 stressLength  190  190  200  200  0  0  220  240

espeak-ng-data/lang/bat/lt

 name Lithuanian
 language lt
+stressRule 2

espeak-ng-data/lang/bat/lv

 tone 150 220 450 255 750 20 3500 255
 stressAmp 12 10 8 8 0 0 18 15
 stressLength 160 140 200 140 0 0 240 180
+stressRule 0

espeak-ng-data/lang/bnt/sw

 language sw
 
 status testing
+stressRule 2

espeak-ng-data/lang/bnt/tn

 language tn
 
 status testing
+stressRule 2

espeak-ng-data/lang/ccs/ka

 name Georgian
 language ka
+stressRule 0

espeak-ng-data/lang/cel/cy

 language cy
 
 intonation 4
+letterVowel w
+letterVowel y
+stressRule 2

espeak-ng-data/lang/cel/ga

 language ga
 
 dictrules 1  // fix for eclipsis
+stressRule 0

espeak-ng-data/lang/cel/gd

 language gd
 
 status testing
+stressRule 0

espeak-ng-data/lang/cus/om

 language om
 
 status testing
+stressRule 2

espeak-ng-data/lang/dra/kn

 
 intonation 2
 //consonants 80
+stressRule 0

espeak-ng-data/lang/dra/ml

 
 intonation 2
 //consonants 80
+stressRule 13

espeak-ng-data/lang/dra/ta

 
 intonation 2
 consonants 80
+stressRule 0

espeak-ng-data/lang/dra/te

 
 intonation 2
 //consonants 80
+stressRule 0

espeak-ng-data/lang/esx/kl

 name Greenlandic
 language kl
 
+stressRule 12

espeak-ng-data/lang/eu

 language eu
 
 status testing
+stressRule 1

espeak-ng-data/lang/gmq/da

 language da
 
 tunes s2 c2 q2 e2
+letterVowel y
+stressRule 0

espeak-ng-data/lang/gmq/is

 name Icelandic
 language is
+letterVowel y
+stressRule 0

espeak-ng-data/lang/gmq/nb

 dictionary no
 
 intonation 4
+letterVowel y
+stressRule 0

espeak-ng-data/lang/gmq/sv

 name Swedish
 language sv
+letterVowel y
+stressRule 0

espeak-ng-data/lang/gmw/af

 
 roughness 0
 pitch 63 120
+letterVowel y
+stressRule 0

espeak-ng-data/lang/gmw/de

 name German
 language de
+
+letterVowel y
+stressRule 0

espeak-ng-data/lang/gmw/en

 status mature
 
 tunes s1 c1 q1 e1
+stressRule 0

espeak-ng-data/lang/gmw/nl

 name Dutch
 language nl
+letterVowel y
+stressRule 0

espeak-ng-data/lang/grk/el

 name Greek
 language el
+stressRule 2

espeak-ng-data/lang/grk/grc

 language grc
 
 stressLength 170 170  190 190  0 0  230 240
+stressRule 2
 dictrules 1
 words 3

espeak-ng-data/lang/inc/as

 language as
 
 status testing
+
+stressRule 0

espeak-ng-data/lang/inc/bn

 name Bengali
 language bn
+stressRule 0

espeak-ng-data/lang/inc/bpy

 name Bishnupriya Manipuri
 language bpy
+stressRule 0

espeak-ng-data/lang/inc/gu

 language gu
 
 status testing
+stressRule 2

espeak-ng-data/lang/inc/hi

 name Hindi
 language hi
+stressRule 6

espeak-ng-data/lang/inc/mr

 language mr
 
 status testing
+stressRule 6 

espeak-ng-data/lang/inc/ne

 language ne
 
 dictrules 1
+stressRule 6

espeak-ng-data/lang/inc/or

 language or
 
 status testing
+stressRule 6

espeak-ng-data/lang/inc/pa

 name Punjabi
 language pa
+stressRule 6

espeak-ng-data/lang/inc/si

 status testing
 
 intonation 2
+stressRule 0

espeak-ng-data/lang/inc/ur

 maintainer Ejaz Shah <[email protected]>
 status testing
 
-stressrule 6
+stressRule 6
 

espeak-ng-data/lang/ine/hy

 name Armenian (East Armenia)
 language hy
 language hy-arevela
+stressRule 3

espeak-ng-data/lang/ine/sq

 name Albanian
 language sq
-
+letterVowel y
+stressRule 3
 // add this line to remove 'ë' at the end of words
 // replace 00  @/  NULL

espeak-ng-data/lang/ira/ku

 language ku
 
 //words 1 48
-
+stressRule 7

espeak-ng-data/lang/itc/la

 name Latin
 language la
-stressrule 2 33 0 2
+stressRule 2 33 0 2
 // rule=penultimate
 // flags=0100001 (no automatic secondary stress + don't stres monosyllables)
 // unstressed_wd1=0

espeak-ng-data/lang/ko

 pitch 80 118
 intonation 2
 
+stressRule 8

espeak-ng-data/lang/poz/id

 
 stressLength 160 200  180 180  0 0  220 240
 stressAmp    16  18   18  18   0 0  22  21
-
+stressRule 2
 consonants 80 80

espeak-ng-data/lang/poz/ms

 
 stressLength 160 200  180 180  0 0  220 240
 stressAmp    16  18   18  18   0 0  22  21
+stressRule 2
 intonation	3	// Less intonation, and comma does not raise the pitch.
 
 // Nuance - Peninsula Malaysia
 				// (only the last phoneme of a word, only in unstressed syllables)
 				
 consonants 80 80
+

espeak-ng-data/lang/roa/an

 name Aragonese
 language an
+stressRule 2

espeak-ng-data/lang/roa/ca

 name Catalan
 language ca
+stressRule 2

espeak-ng-data/lang/roa/es

 language es
 dictrules 1
 tunes s6 c6 q6 e6
+stressRule 2

espeak-ng-data/lang/roa/fr

 
 dictrules 1
 tunes s3 c3 q3 e3
+letterVowel y
+stressRule 3

espeak-ng-data/lang/roa/fr-BE

 
 dictrules 2
 tunes s3 c3 q3 e3
-
+letterVowel y
 

espeak-ng-data/lang/roa/fr-CH

 
 dictrules 3
 tunes s3 c3 q3 e3
+letterVowel y

espeak-ng-data/lang/roa/it

 status mature
 
 tunes s4 c4 q4 e4
+letterVowel y
+stressRule 2

espeak-ng-data/lang/roa/pap

 status testing
 
 phonemes base2
-
+stressRule 3

espeak-ng-data/lang/roa/pt

 
 dictrules 1
 intonation 2
+letterVowel y
+stressRule 3

espeak-ng-data/lang/roa/pt-BR

 
 dictrules 2
 stressLength 200 115 230 230 0 0 250 270
-
+stressRule 3

espeak-ng-data/lang/sai/gn

 language gn
 dictrules 1
 words 0 1
+stressRule 3

espeak-ng-data/lang/sem/am

 language am
 
 status testing
+stressRule 0

espeak-ng-data/lang/sem/mt

 language mt
 
 status testing
+stressRule 2

espeak-ng-data/lang/trk/az

 language az
 
 status testing
+stressRule 7

espeak-ng-data/lang/trk/tr

 name Turkish
 language tr
+stressRule 7

espeak-ng-data/lang/trk/tt

 name Tatar
 language tt
+stressRule 3

espeak-ng-data/lang/urj/fi

 name Finnish
 language fi
+letterVowel y
+stressRule 0

espeak-ng-data/lang/urj/hu

 language hu
 option bracket 0 0
 pitch 81 117
-
-
+letterVowel y
+stressRule 0

espeak-ng-data/lang/zle/ru

 replace 03 a a#
 
 dict_min  20000
+stressRule 5

espeak-ng-data/lang/zls/bg

 
 stressAmp 13 12 17 17 20 22 22 21 
 stressLength 180 170  200 200  200 200  210 220
+stressRule 2
+letterVowel 0x2a
+

espeak-ng-data/lang/zls/bs

 
 stressAdd 10 10 0 0 0 0 -30 -30
 dictrules 3 4
+letterVowel y
+letterVowel r
+stressRule 0

espeak-ng-data/lang/zls/hr

 
 stressAdd 10 10 0 0 0 0 -30 -30
 dictrules 1
+letterVowel y
+letterVowel r
+stressRule 0

espeak-ng-data/lang/zls/mk

 name Macedonian
 language mk
+stressRule 4

espeak-ng-data/lang/zls/sl

 language sl
 
 status testing
+stressRule 2

espeak-ng-data/lang/zls/sr

 
 stressAdd 10 10 0 0 0 0 -30 -30
 dictrules 2 4
+
+letterVowel y
+letterVowel r
+stressRule 0

espeak-ng-data/lang/zlw/cs

 name Czech
 language cs
+letterVowel y
+letterVowel r
+stressRule 0

espeak-ng-data/lang/zlw/pl

 language pl
 
 intonation 2
+letterVowel y
+stressRule 2

espeak-ng-data/lang/zlw/sk

 name Slovak
 language sk
+letterVowel y
+letterVowel r
+stressRule 0

src/libespeak-ng/tr_languages.c

 
 static void Translator_Russian(Translator *tr);
 
-static void SetLetterVowel(Translator *tr, int c)
-{
-	tr->letter_bits[c] = (tr->letter_bits[c] & 0x40) | 0x81; // keep value for group 6 (front vowels e,i,y)
-}
-
 static void ResetLetterBits(Translator *tr, int groups)
 {
 	// Clear all the specified groups
 		static const short stress_lengths_af[8] = { 170, 140, 220, 220,  0, 0, 250, 270 };
 		SetupTranslator(tr, stress_lengths_af, NULL);
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.vowel_pause = 0x30;
 		tr->langopts.param[LOPT_DIERESES] = 1;
 		tr->langopts.param[LOPT_PREFIXES] = 1;
-		SetLetterVowel(tr, 'y'); // add 'y' to vowels
 
 		tr->langopts.numbers = NUM_SWAP_TENS | NUM_HUNDRED_AND | NUM_SINGLE_AND | NUM_ROMAN | NUM_1900;
 		tr->langopts.accents = 1;
 	{
 		SetupTranslator(tr, stress_lengths_fr, stress_amps_fr);
 		tr->letter_bits_offset = OFFSET_ETHIOPIC;
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.stress_flags = S_NO_AUTO_2 | S_FINAL_DIM; // don't use secondary stress
 		tr->langopts.length_mods0 = tr->langopts.length_mods;  // don't lengthen vowels in the last syllable
 		tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1;           // disable check for unpronouncable words
 	case L('b', 'g'): // Bulgarian
 	{
 		SetCyrillicLetters(tr);
-		SetLetterVowel(tr, 0x2a);
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_5;
 		tr->langopts.param[LOPT_UNPRONOUNCABLE] = 0x432; // [v]  don't count this character at start of word
 		tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 0x107; // devoice at end of word, and change voicing to match a following consonant (except v)
 		tr->langopts.param[LOPT_REDUCE] = 2;
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_ALLOW_SPACE | NUM_OMIT_1_HUNDRED | NUM_HUNDRED_AND | NUM_AND_UNITS | NUM_SINGLE_AND | NUM_ROMAN | NUM_ROMAN_ORDINAL | NUM_ROMAN_CAPITALS;
 		tr->langopts.thousands_sep = ' '; // don't allow dot as thousands separator
 	}
 		SetupTranslator(tr, stress_lengths_bn, stress_amps_bn);
 		tr->langopts.length_mods0 = tr->langopts.length_mods; // don't lengthen vowels in the last syllable
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.stress_flags =  S_MID_DIM | S_FINAL_DIM; // use 'diminished' for unstressed final syllable
 		tr->letter_bits_offset = OFFSET_BENGALI;
 		SetIndicLetters(tr); // call this after setting OFFSET_BENGALI
 		SetupTranslator(tr, stress_lengths_cy, stress_amps_cy);
 
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_14;
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 
 		// 'diminished' is an unstressed final syllable
 		tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2;
 
 		tr->langopts.numbers = NUM_OMIT_1_HUNDRED;
 
-		SetLetterVowel(tr, 'w'); // add letter to vowels and remove from consonants
-		SetLetterVowel(tr, 'y');
 	}
 		break;
 	case L('d', 'a'): // Danish
 		static const short stress_lengths_da[8] = { 160, 140, 200, 200, 0, 0, 220, 230 };
 		SetupTranslator(tr, stress_lengths_da, NULL);
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.param[LOPT_PREFIXES] = 1;
-		SetLetterVowel(tr, 'y');
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SWAP_TENS | NUM_HUNDRED_AND | NUM_OMIT_1_HUNDRED | NUM_ORDINAL_DOT | NUM_1900 | NUM_ROMAN | NUM_ROMAN_CAPITALS | NUM_ROMAN_ORDINAL;
 	}
 		break;
 		static const short stress_lengths_de[8] = { 150, 130, 200, 200,  0, 0, 270, 270 };
 		static const unsigned char stress_amps_de[] = { 20, 20, 20, 20, 20, 22, 22, 20 };
 		SetupTranslator(tr, stress_lengths_de, stress_amps_de);
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.word_gap = 0x8; // don't use linking phonemes
 		tr->langopts.vowel_pause = 0x30;
 		tr->langopts.param[LOPT_PREFIXES] = 1;
 		tr->langopts.param[LOPT_LONG_VOWEL_THRESHOLD] = 175/2;
 
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SWAP_TENS | NUM_ALLOW_SPACE | NUM_ORDINAL_DOT | NUM_ROMAN;
-		SetLetterVowel(tr, 'y');
 		tr->langopts.param[LOPT_UNPRONOUNCABLE] = 2; // use de_rules for unpronouncable rules
 	}
 		break;
 		static const short stress_lengths_en[8] = { 182, 140, 220, 220, 0, 0, 248, 275 };
 		SetupTranslator(tr, stress_lengths_en, NULL);
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.stress_flags = 0x08;
 		tr->langopts.numbers = NUM_HUNDRED_AND | NUM_ROMAN | NUM_1900;
 		tr->langopts.param[LOPT_COMBINE_WORDS] = 2; // allow "mc" to cmbine with the following word
 		SetLetterBits(tr, LETTERGP_Y, el_fvowels); // front vowels: ε η ι υ _
 
 		tr->langopts.length_mods0 = tr->langopts.length_mods; // don't lengthen vowels in the last syllable
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.stress_flags = S_FINAL_DIM_ONLY; // mark unstressed final syllables as diminished
 		tr->langopts.unstressed_wd1 = 0;
 		tr->langopts.unstressed_wd2 = 2;
 		tr->char_plus_apostrophe = eo_char_apostrophe;
 
 		tr->langopts.vowel_pause = 2;
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2;
 		tr->langopts.unstressed_wd2 = 2;
 
 		SetupTranslator(tr, stress_lengths_es, stress_amps_es);
 
 		tr->langopts.length_mods0 = tr->langopts.length_mods; // don't lengthen vowels in the last syllable
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 
 		// stress last syllable if it doesn't end in vowel or "s" or "n"
 		// 'diminished' is an unstressed final syllable
 			tr->langopts.roman_suffix = utf8_ordinal;
 		} else if (name2 == L_pap) {
 			// stress last syllable unless word ends with a vowel
-			tr->langopts.stress_rule = STRESSPOSN_1R;
 			tr->langopts.stress_flags = S_FINAL_VOWEL_UNSTRESSED | S_FINAL_DIM_ONLY | S_FINAL_NO_2 | S_NO_AUTO_2;
 		} else
 			tr->langopts.param[LOPT_UNPRONOUNCABLE] = 2; // use es_rules for unpronouncable rules
 		static const short stress_lengths_eu[8] = { 200, 200,  200, 200,  0, 0,  210, 230 }; // very weak stress
 		static const unsigned char stress_amps_eu[8] = { 16, 16, 18, 18, 18, 18, 18, 18 };
 		SetupTranslator(tr, stress_lengths_eu, stress_amps_eu);
-		tr->langopts.stress_rule = STRESSPOSN_2L; // ?? second syllable, but not on a word-final vowel
 		tr->langopts.stress_flags = S_FINAL_VOWEL_UNSTRESSED;
 		tr->langopts.param[LOPT_SUFFIX] = 1;
 		tr->langopts.numbers = NUM_SINGLE_STRESS | NUM_DECIMAL_COMMA | NUM_HUNDRED_AND | NUM_OMIT_1_HUNDRED | NUM_OMIT_1_THOUSAND | NUM_VIGESIMAL;
 
 		SetupTranslator(tr, stress_lengths_fi, stress_amps_fi);
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.stress_flags = S_FINAL_DIM_ONLY | S_FINAL_NO_2 | S_2_TO_HEAVY; // move secondary stress from light to a following heavy syllable
 		tr->langopts.param[LOPT_IT_DOUBLING] = 1;
 		tr->langopts.long_stop = 130;
 
 		tr->langopts.numbers = NUM_DECIMAL_COMMA + NUM_ALLOW_SPACE;
-		SetLetterVowel(tr, 'y');
 		tr->langopts.spelling_stress = 1;
 		tr->langopts.intonation_group = 3; // less intonation, don't raise pitch at comma
 	}
 	case L('f', 'r'): // french
 	{
 		SetupTranslator(tr, stress_lengths_fr, stress_amps_fr);
-		tr->langopts.stress_rule = STRESSPOSN_1R; // stress on final syllable
 		tr->langopts.stress_flags = S_NO_AUTO_2 | S_FINAL_DIM; // don't use secondary stress
 		tr->langopts.param[LOPT_IT_LENGTHEN] = 1; // remove lengthen indicator from unstressed syllables
 		tr->langopts.length_mods0 = tr->langopts.length_mods; // don't lengthen vowels in the last syllable
 		tr->langopts.accents = 2; // Say "Capital" after the letter.
 
 		tr->langopts.numbers = NUM_SINGLE_STRESS | NUM_DECIMAL_COMMA | NUM_ALLOW_SPACE | NUM_OMIT_1_HUNDRED | NUM_NOPAUSE | NUM_ROMAN | NUM_ROMAN_CAPITALS | NUM_ROMAN_AFTER | NUM_VIGESIMAL | NUM_DFRACTION_4;
-		SetLetterVowel(tr, 'y');
 	}
 		break;
 	case L('g', 'a'): // irish
 		tr->encoding = ESPEAKNG_ENCODING_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, excluding final syllable
 		tr->langopts.stress_flags =  S_MID_DIM | S_FINAL_DIM; // use 'diminished' for unstressed final syllable
 		tr->langopts.numbers = NUM_SWAP_TENS;
 		tr->langopts.break_numbers = 0x14aa8; // for languages which have numbers for 100,000 and 100,00,000, eg Hindi
 		else if (name2 == L('g', 'u')) {
 			SetupTranslator(tr, stress_lengths_equal, stress_amps_equal);
 			tr->letter_bits_offset = OFFSET_GUJARATI;
-			tr->langopts.stress_rule = STRESSPOSN_2R;
 		} else if (name2 == L('n', 'e')) {
 			SetupTranslator(tr, stress_lengths_equal, stress_amps_equal);
 			tr->langopts.break_numbers = 0x2aaaa8;
 			SetupTranslator(tr, stress_lengths_hr, stress_amps_hr);
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_2;
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.stress_flags = S_FINAL_NO_2;
 		tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 0x3;
 		tr->langopts.max_initial_consonants = 5;
 		tr->langopts.replace_chars = replace_cyrillic_latin;
 		tr->langopts.our_alphabet = OFFSET_CYRILLIC; // don't say "cyrillic" before letter names
 
-		SetLetterVowel(tr, 'y');
-		SetLetterVowel(tr, 'r');
 	}
 		break;
 	case L('h', 't'): // Haitian Creole
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_2;
 
 		tr->langopts.vowel_pause = 0x20;
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.stress_flags = S_FINAL_DIM_ONLY | S_FINAL_NO_2 | S_NO_AUTO_2 | 0x8000 | S_HYPEN_UNSTRESS;
 		tr->langopts.unstressed_wd1 = 2;
 		tr->langopts.param[LOPT_IT_DOUBLING] = 1;
 		tr->langopts.decimal_sep = ',';
 		tr->langopts.max_roman = 899;
 		tr->langopts.min_roman = 1;
-		SetLetterVowel(tr, 'y');
 		tr->langopts.spelling_stress = 1;
 		SetLengthMods(tr, 3); // all equal
 	}
 		static const char hy_consonants2[] = { 0x45, 0 };
 
 		SetupTranslator(tr, stress_lengths_hy, NULL);
-		tr->langopts.stress_rule = STRESSPOSN_1R; // default stress on final syllable
 
 		tr->letter_bits_offset = OFFSET_ARMENIAN;
 		memset(tr->letter_bits, 0, sizeof(tr->letter_bits));
 		static const unsigned char stress_amps_id[8] = { 16, 18, 18, 18, 20, 22, 22, 21 };
 
 		SetupTranslator(tr, stress_lengths_id, stress_amps_id);
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_ALLOW_SPACE | NUM_ROMAN;
 		tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2;
 		tr->langopts.accents = 2; // "capital" after letter name
 		static const wchar_t is_lettergroup_B[] = { 'c', 'f', 'h', 'k', 'p', 't', 'x', 0xfe, 0 }; // voiceless conants, including 'þ'  ?? 's'
 
 		SetupTranslator(tr, stress_lengths_is, NULL);
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.stress_flags = S_FINAL_NO_2;
 		tr->langopts.param[LOPT_IT_LENGTHEN] = 0x11; // remove lengthen indicator from unstressed vowels
 		tr->langopts.param[LOPT_REDUCE] = 2;
 		SetLetterBits(tr, 4, "kpst"); // Letter group F
 		SetLetterBits(tr, 3, "jvr"); // Letter group H
 		tr->letter_groups[1] = is_lettergroup_B;
-		SetLetterVowel(tr, 'y');
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SINGLE_AND | NUM_HUNDRED_AND | NUM_AND_UNITS | NUM_1900;
 		tr->langopts.numbers2 = 0x2;
 	}
 		static const unsigned char stress_amps_it[8] = { 17, 15, 18, 16, 20, 22, 22, 22 };
 		SetupTranslator(tr, stress_lengths_it, stress_amps_it);
 		tr->langopts.length_mods0 = tr->langopts.length_mods; // don't lengthen vowels in the last syllable
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.stress_flags = S_NO_AUTO_2 | S_FINAL_DIM_ONLY | S_PRIORITY_STRESS;
 		tr->langopts.vowel_pause = 1;
 		tr->langopts.unstressed_wd1 = 0;
 		tr->langopts.numbers2 = NUM2_NO_TEEN_ORDINALS;
 		tr->langopts.roman_suffix = utf8_ordinal;
 		tr->langopts.accents = 2; // Say "Capital" after the letter.
-		SetLetterVowel(tr, 'y');
 	}
 		break;
 	case L_jbo: // Lojban
 		static const wchar_t jbo_punct_within_word[] = { '.', ',', '\'', 0x2c8, 0 }; // allow period and comma within a word, also stress marker (from LOPT_CAPS_IN_WORD)
 
 		SetupTranslator(tr, stress_lengths_jbo, NULL);
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.vowel_pause = 0x20c; // pause before a word which starts with a vowel, or after a word which ends in a consonant
 		tr->punct_within_word = jbo_punct_within_word;
 		tr->langopts.param[LOPT_CAPS_IN_WORD] = 2; // capitals indicate stressed syllables
-		SetLetterVowel(tr, 'y');
 		tr->langopts.max_lengthmod = 368;
 	}
 		break;
 		SetLetterBits(tr, LETTERGP_C, ka_consonants);
 		SetLetterBits(tr, LETTERGP_VOWEL2, ka_vowels);
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.stress_flags = S_FINAL_NO_2;
 		tr->letter_bits_offset = OFFSET_GEORGIAN;
 		tr->langopts.max_initial_consonants = 7;
 	case L('k', 'l'): // Greenlandic
 	{
 		SetupTranslator(tr, stress_lengths_equal, stress_amps_equal);
-		tr->langopts.stress_rule = 12;
 		tr->langopts.stress_flags = S_NO_AUTO_2;
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SWAP_TENS | NUM_HUNDRED_AND | NUM_OMIT_1_HUNDRED | NUM_ORDINAL_DOT | NUM_1900 | NUM_ROMAN | NUM_ROMAN_CAPITALS | NUM_ROMAN_ORDINAL;
 	}
 		SetLetterBits(tr, LETTERGP_Y, ko_ivowels);
 		SetLetterBits(tr, LETTERGP_G, (const char *)ko_voiced);
 
-		tr->langopts.stress_rule = 8; // ?? 1st syllable if it is heavy, else 2nd syllable
 		tr->langopts.param[LOPT_UNPRONOUNCABLE] = 1; // disable check for unpronouncable words
 		tr->langopts.numbers = NUM_OMIT_1_HUNDRED;
 		tr->langopts.numbers2 = NUM2_MYRIADS;
 		SetupTranslator(tr, stress_lengths_ku, stress_amps_ku);
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_9;
 
-		tr->langopts.stress_rule = 7; // stress on the last syllable, before any explicitly unstressed syllable
-
 		tr->langopts.numbers = NUM_HUNDRED_AND | NUM_AND_UNITS | NUM_OMIT_1_HUNDRED | NUM_AND_HUNDRED;
 		tr->langopts.max_initial_consonants = 2;
 	}
 	case L('l', 'a'): // Latin
 	{
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_4; // includes a,e,i,o,u-macron
-		tr->langopts.stress_rule = STRESSPOSN_2R;
-		tr->langopts.stress_flags = S_NO_AUTO_2;
-		tr->langopts.unstressed_wd1 = 0;
-		tr->langopts.unstressed_wd2 = 2;
 		tr->langopts.param[LOPT_DIERESES] = 1;
 		tr->langopts.numbers = NUM_ROMAN;
 		tr->langopts.max_roman = 5000;
 	case L('l', 't'): // Lithuanian
 	{
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_4;
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.stress_flags = S_NO_AUTO_2;
 		tr->langopts.unstressed_wd1 = 0;
 		tr->langopts.unstressed_wd2 = 2;
 
 		SetupTranslator(tr, stress_lengths_lv, stress_amps_lv);
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.spelling_stress = 1;
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_4;
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_OMIT_1_HUNDRED | NUM_DFRACTION_4 | NUM_ORDINAL_DOT;
 		tr->letter_groups[0] = tr->letter_groups[7] = vowels_cyrillic;
 		tr->letter_bits_offset = OFFSET_CYRILLIC;
 
-		tr->langopts.stress_rule = STRESSPOSN_3R; // antipenultimate
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_AND_UNITS | NUM_OMIT_1_HUNDRED | NUM_OMIT_1_THOUSAND | NUM_DFRACTION_2;
 		tr->langopts.numbers2 = 0x8a; // variant numbers before thousands,milliards
 	}
 	{
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_3;
 		tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 0x100; // devoice at end of word
-		tr->langopts.stress_rule = STRESSPOSN_2R; // penultimate
 		tr->langopts.numbers = 1;
 	}
 		break;
 	{
 		static const short stress_lengths_nl[8] = { 160, 135, 210, 210,  0, 0, 260, 280 };
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.vowel_pause = 0x30; // ??
 		tr->langopts.param[LOPT_DIERESES] = 1;
 		tr->langopts.param[LOPT_PREFIXES] = 1;
 		tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 0x100; // devoice at end of word
-		SetLetterVowel(tr, 'y');
 
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_SWAP_TENS | NUM_OMIT_1_HUNDRED | NUM_OMIT_1_THOUSAND | NUM_ALLOW_SPACE | NUM_1900 | NUM_ORDINAL_DOT;
 		tr->langopts.ordinal_indicator = "e";
 		static const short stress_lengths_no[8] = { 160, 140, 200, 200, 0, 0, 220, 230 };
 
 		SetupTranslator(tr, stress_lengths_no, NULL);
-		tr->langopts.stress_rule = STRESSPOSN_1L;
-		SetLetterVowel(tr, 'y');
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_HUNDRED_AND | NUM_ALLOW_SPACE | NUM_1900 | NUM_ORDINAL_DOT;
 	}
 		break;
 		static const short stress_lengths_om[8] = { 200, 200, 200, 200, 0, 0, 200, 200 };
 
 		SetupTranslator(tr, stress_lengths_om, stress_amps_om);
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.stress_flags = S_FINAL_DIM_ONLY | S_FINAL_NO_2 | 0x80000;
 		tr->langopts.numbers = NUM_OMIT_1_HUNDRED | NUM_HUNDRED_AND;
 		tr->langopts.numbers2 = 0x200; // say "thousands" before its number
 		SetupTranslator(tr, stress_lengths_pl, stress_amps_pl);
 
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_2;
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.stress_flags = S_FINAL_DIM_ONLY; // mark unstressed final syllables as diminished
 		tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 0x9;
 		tr->langopts.max_initial_consonants = 7; // for example: wchrzczony :)
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_ALLOW_SPACE | NUM_DFRACTION_2;
 		tr->langopts.numbers2 = NUM2_THOUSANDS_VAR3;
 		tr->langopts.param[LOPT_COMBINE_WORDS] = 4 + 0x100; // combine 'nie' (marked with $alt2) with some 1-syllable (and 2-syllable) words (marked with $alt)
-		SetLetterVowel(tr, 'y');
 	}
 		break;
 	case L('p', 't'): // Portuguese
 		SetupTranslator(tr, stress_lengths_pt, stress_amps_pt);
 		tr->langopts.length_mods0 = tr->langopts.length_mods; // don't lengthen vowels in the last syllable
 
-		tr->langopts.stress_rule = STRESSPOSN_1R; // stress on final syllable
 		tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2 | S_INITIAL_2 | S_PRIORITY_STRESS;
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_DFRACTION_2 | NUM_HUNDRED_AND | NUM_AND_UNITS | NUM_ROMAN_CAPITALS;
 		tr->langopts.numbers2 = NUM2_MULTIPLE_ORDINAL | NUM2_NO_TEEN_ORDINALS | NUM2_ORDINAL_NO_AND;
 		tr->langopts.max_roman = 5000;
-		SetLetterVowel(tr, 'y');
 		ResetLetterBits(tr, 0x2);
 		SetLetterBits(tr, 1, "bcdfgjkmnpqstvxz"); // B  hard consonants, excluding h,l,r,w,y
 		tr->langopts.param[LOPT_ALT] = 2; // call ApplySpecialAttributes2() if a word has $alt or $alt2
 
 		SetupTranslator(tr, stress_lengths_ro, stress_amps_ro);
 
-		tr->langopts.stress_rule = STRESSPOSN_1R;
 		tr->langopts.stress_flags = S_FINAL_VOWEL_UNSTRESSED | S_FINAL_DIM_ONLY;
 
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_2;
 		SetupTranslator(tr, stress_lengths_sk, stress_amps_sk);
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_2;
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.stress_flags = S_FINAL_DIM_ONLY | S_FINAL_NO_2;
 		tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 0x3;
 		tr->langopts.max_initial_consonants = 5;
 		if (name2 == L('c', 's'))
 			tr->langopts.numbers2 = 0x108; // variant numbers before milliards
 
-		SetLetterVowel(tr, 'y');
-		SetLetterVowel(tr, 'r');
 		ResetLetterBits(tr, 0x20);
 		SetLetterBits(tr, 5, sk_voiced);
 	}
 		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 = STRESSPOSN_1L;
 		tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2;
 		tr->langopts.spelling_stress = 1;
 
 		break;
 	case L('s', 'l'): // Slovenian
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_2;
-		tr->langopts.stress_rule = STRESSPOSN_2R; // Temporary
 		tr->langopts.stress_flags = S_NO_AUTO_2;
 		tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 0x103;
 		tr->langopts.param[LOPT_UNPRONOUNCABLE] = 0x76; // [v]  don't count this character at start of word
 
 		SetupTranslator(tr, stress_lengths_sq, stress_amps_sq);
 
-		tr->langopts.stress_rule = STRESSPOSN_1R;
 		tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2 | S_FINAL_VOWEL_UNSTRESSED;
-		SetLetterVowel(tr, 'y');
 		tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_HUNDRED_AND | NUM_AND_UNITS | NUM_DFRACTION_4;
 		tr->langopts.accents = 2; // "capital" after letter name
 	}
 		static const short stress_lengths_sv[8] = { 160, 135, 220, 220, 0, 0, 250, 280 };
 		SetupTranslator(tr, stress_lengths_sv, stress_amps_sv);
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
-		SetLetterVowel(tr, 'y');
 		tr->langopts.numbers = NUM_SINGLE_STRESS | NUM_DECIMAL_COMMA | NUM_ALLOW_SPACE | NUM_1900;
 		tr->langopts.accents = 1;
 	}
 		tr->langopts.length_mods0 = tr->langopts.length_mods; // don't lengthen vowels in the last syllable
 
 		tr->langopts.vowel_pause = 1;
-		tr->langopts.stress_rule = STRESSPOSN_2R;
 		tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2;
 		tr->langopts.max_initial_consonants = 4; // for example: mwngi
 
 		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;
 		tr->langopts.stress_flags =  S_FINAL_DIM_ONLY | S_FINAL_NO_2; // use 'diminished' for unstressed final syllable
 		tr->langopts.spelling_stress = 1;
 		tr->langopts.break_numbers = 0x14a8; // 1000, 100,000  10,000,000
 			tr->letter_bits_offset = OFFSET_MALAYALAM;
 			tr->langopts.numbers = NUM_OMIT_1_THOUSAND | NUM_OMIT_1_HUNDRED;
 			tr->langopts.numbers2 = NUM2_OMIT_1_HUNDRED_ONLY;
-			tr->langopts.stress_rule = 13; // 1st syllable, unless 1st vowel is short and 2nd is long
 		} else if (name2 == L('k', 'n')) {
 			tr->letter_bits_offset = OFFSET_KANNADA;
 			tr->langopts.numbers = 0x1;
 		SetupTranslator(tr, stress_lengths_tr, stress_amps_tr);
 		tr->encoding = ESPEAKNG_ENCODING_ISO_8859_9;
 
-		tr->langopts.stress_rule = 7; // stress on the last syllable, before any explicitly unstressed syllable
 		tr->langopts.stress_flags = S_NO_AUTO_2; // no automatic secondary stress
 		tr->langopts.dotless_i = 1;
 		tr->langopts.param[LOPT_SUFFIX] = 1;
 	{
 		SetCyrillicLetters(tr);
 		SetupTranslator(tr, stress_lengths_fr, stress_amps_fr);
-		tr->langopts.stress_rule = STRESSPOSN_1R; // stress on final syllable
 		tr->langopts.stress_flags = S_NO_AUTO_2; // no automatic secondary stress
 		tr->langopts.numbers = NUM_SINGLE_STRESS | NUM_DECIMAL_COMMA | NUM_OMIT_1_HUNDRED | NUM_OMIT_1_THOUSAND | NUM_DFRACTION_4;
 	}
 		SetupTranslator(tr, stress_lengths_vi, stress_amps_vi);
 		tr->langopts.length_mods0 = tr->langopts.length_mods; // don't lengthen vowels in the last syllable
 
-		tr->langopts.stress_rule = STRESSPOSN_1L;
 		tr->langopts.word_gap = 0x21; // length of a final vowel is less dependent on the next consonant, don't merge consonant with next word
 		tr->letter_groups[0] = tr->letter_groups[7] = vowels_vi;
 		tr->langopts.tone_language = 1; // Tone language, use  CalcPitches_Tone() rather than CalcPitches()
 	tr->langopts.param[LOPT_UNPRONOUNCABLE] = 0x432; // [v]  don't count this character at start of word
 	tr->langopts.param[LOPT_REGRESSIVE_VOICING] = 1;
 	tr->langopts.param[LOPT_REDUCE] = 2;
-	tr->langopts.stress_rule = 5;
 	tr->langopts.stress_flags = S_NO_AUTO_2;
 
 	tr->langopts.numbers = NUM_DECIMAL_COMMA | NUM_OMIT_1_HUNDRED;

src/libespeak-ng/voices.c

 
 	// these need a phoneme table to have been specified
 	V_REPLACE,
-	V_CONSONANTS
+	V_CONSONANTS,
+
+	// these are alpha features that need to be tested and categorized
+	V_LETTER_VOWEL
+
 };
 
 static MNEM_TAB options_tab[] = {
 	{ "intonation",   V_INTONATION },
 	{ "tunes",        V_TUNES },
 	{ "dictrules",    V_DICTRULES },
-	{ "stressrule",   V_STRESSRULE },
+	{ "stressRule",   V_STRESSRULE },
 	{ "stressopt",    V_STRESSOPT },
 	{ "replace",      V_REPLACE },
 	{ "words",        V_WORDGAP },
 	{ "l_length_mods",    0x100+LOPT_LENGTH_MODS },
 	{ "apostrophe",       0x100+LOPT_APOSTROPHE },
 
+	// these are alpha features that need to be tested and categorized
+	{ "letterVowel", V_LETTER_VOWEL },
 	{ NULL, 0 }
 };
 
 		case V_MAINTAINER:
 		case V_STATUS:
 			break;
+
+		case V_LETTER_VOWEL: {
+			char str[5] = "";
+			char c = '0';
+			char *endptr = NULL;
+			sscanf(p, "%s", str);
+			// assume a hex value if string starts with "0x"
+			if (str[0] == '0' && str[1] == 'x') {
+				c = strtoul(str, &endptr, 16);
+				if (errno == ERANGE)
+					fprintf(stderr, "letterVowel out of range.\n");
+			}
+			else  {// otherwise, assume a single letter
+				c = str[0];
+				if (c < 97 || c > 122) // valid values are a-z, ascii 97-122
+					fprintf(stderr, "letterVowel out of range.\n");
+			}
+ 			new_translator->letter_bits[c] = (new_translator->letter_bits[c] & 0x40) | 0x81; // keep value for group 6 (front vowels e,i,y)
+			break;
+			}
+
 		default:
 			if ((key & 0xff00) == 0x100) {
 				if (langopts)