mjohnson/espeak-ng
1
0
Fork 0
mirror of https://github.com/rhasspy/espeak-ng.git synced 2026-10-31 05:35:35 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
8a6e17ac7c25334ca4de7e18d29a1c3f6b338fe7
espeak-ng/src/translate.h
T
jonsd 8a6e17ac7c [1.25.02] Added translation for mbrola "gr2" Greek voice. 
Filtered ustop/t_short.wav to remove high frequencies.
Reduced formant glide for phoneme [c].


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@18 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-05-21 17:10:30 +00:00

518 lines
21 KiB
C++
Executable File
Raw Blame History

/***************************************************************************
* Copyright (C) 2005,2006 by Jonathan Duddington *
* jonsd@users.sourceforge.net *
* *
* This program is free software; you can redistribute it and/or modify *
* it under the terms of the GNU General Public License as published by *
* the Free Software Foundation; either version 2 of the License, or *
* (at your option) any later version. *
* *
* This program is distributed in the hope that it will be useful, *
* but WITHOUT ANY WARRANTY; without even the implied warranty of *
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the *
* GNU General Public License for more details. *
* *
* You should have received a copy of the GNU General Public License *
* along with this program; if not, write to the *
* Free Software Foundation, Inc., *
* 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. *
***************************************************************************/
#define L(c1,c2) (c1<<8)+c2 // combine two characters into an integer for translator name
#define CTRL_EMBEDDED 0x01 // control character at the start of an embedded command
#define REPLACED_E 'E' // 'e' replaced by silent e
#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_RULE_GROUP2 120 // max num of two-letter rule chains
#define N_HASH_DICT 1024
#define N_CHARSETS 20
/* flags from word dictionary */
// bits 0-3 stressed syllable, 7=unstressed
#define FLAG_SKIPWORDS 0xe0 /* bits 5,6,7 number of words to skip */
#define FLAG_SKIPWORDS_1 0x20
#define FLAG_PREPAUSE 0x100
#define FLAG_ONLY 0x200
#define BITNUM_FLAG_ONLY 9 // bit 9 is set
#define BITNUM_FLAG_ONLY_S 10 // bit 10 is set
#define FLAG_ONLY_S 0x400
#define FLAG_STRESS_END 0x800 /* full stress if at end of clause */
#define FLAG_STRESS_END2 0x1000 /* full stress if at end of clause, or only followed by unstressed */
#define FLAG_UNSTRESS_END 0x2000 /* reduce stress at end of clause */
#define FLAG_ATEND 0x4000 /* use this pronunciation if at end of clause */
#define FLAG_CAPITAL 0x8000 /* pronunciation if initial letter is upper case */
#define FLAG_DOT 0x10000 /* ignore '.' after word (abbreviation) */
#define FLAG_ABBREV 0x20000 // spell as letters, even with a vowel, OR use specified pronunciation rather than split into letters
#define FLAG_STEM 0x40000 // must have a suffix
#define FLAG_DOUBLING 0x80000 // doubles the following consonant
#define FLAG_ALT_TRANS 0x100000 // language specific
#define FLAG_ALT2_TRANS 0x200000 // language specific
#define FLAG_VERBF 0x400000 /* verb follows */
#define FLAG_VERBSF 0x800000 /* verb follows, may have -s suffix */
#define FLAG_NOUNF 0x1000000 /* noun follows */
#define FLAG_VERB 0x2000000 /* pronunciation for verb */
#define FLAG_PAST 0x4000000 /* pronunciation for past tense */
#define FLAG_PASTF 0x8000000 /* past tense follows */
#define FLAG_VERB_EXT 0x10000000 /* extend the 'verb follows' */
#define FLAG_PAUSE1 0x40000000 // shorter prepause
#define FLAG_FOUND 0x80000000 /* pronunciation was found in the dictionary list */
// wordflags, flags in source word
#define FLAG_ALL_UPPER 0x1 /* no lower case letters in the word */
#define FLAG_FIRST_UPPER 0x2 /* first letter is upper case */
#define FLAG_HAS_PLURAL 0x4 /* upper-case word with s or 's lower-case ending */
#define FLAG_PHONEMES 0x8 /* word is phonemes */
#define FLAG_LAST_WORD 0x10 /* last word in clause */
#define FLAG_STRESSED_WORD 0x20 /* this word has explicit stress */
#define FLAG_EMBEDDED 0x40 /* word is preceded by embedded commands */
#define FLAG_HYPHEN 0x80
#define FLAG_DONT_SWITCH_TRANSLATOR 0x1000
#define FLAG_SUFFIX_REMOVED 0x2000
#define FLAG_HYPHEN_AFTER 0x4000
// prefix/suffix flags (bits 8 to 14, bits 16 to 22) don't use 0x8000, 0x800000
#define SUFX_E 0x0100 // e may have been added
#define SUFX_I 0x0200 // y may have been changed to i
#define SUFX_P 0x0400 // prefix
#define SUFX_V 0x0800 // suffix means use the verb form pronunciation
#define SUFX_D 0x1000 // previous letter may have been doubles
#define SUFX_F 0x2000 // verb follows
#define SUFX_Q 0x4000 // don't retranslate
#define SUFX_T 0x10000 // don't affect the stress position in the stem
#define FLAG_SUFX 0x04
#define FLAG_SUFX_S 0x08
#define FLAG_SUFX_E_ADDED 0x10
// codes in dictionary rules
#define RULE_PRE 1
#define RULE_POST 2
#define RULE_PHONEMES 3
#define RULE_PH_COMMON 4 // At start of rule. Its phoneme string is used by subsequent rules
#define RULE_CONDITION 5 // followed by condition number (byte)
#define RULE_GROUP_START 6
#define RULE_GROUP_END 7
#define RULE_SPACE 32 // ascii space
#define RULE_SYLLABLE 9
#define RULE_STRESSED 10
#define RULE_DOUBLE 11
#define RULE_INC_SCORE 12
#define RULE_DEL_FWD 13
#define RULE_ENDING 14
#define RULE_DIGIT 15 // D digit
#define RULE_NONALPHA 16 // Z non-alpha
#define RULE_LETTER_GROUPS 17 // 17 to 23
#define RULE_LETTER1 17 // A vowels
#define RULE_LETTER2 18 // B 'hard' consonants
#define RULE_LETTER3 19 // C all consonants
#define RULE_LETTER4 20 // H letter group
#define RULE_LETTER5 21 // F letter group
#define RULE_LETTER6 22 // G letter group
#define RULE_LETTER7 23 // Y letter group
#define RULE_NO_SUFFIX 24 // N
#define RULE_NOTVOWEL 25 // K
#define RULE_IFVERB 26 // V
#define RULE_LETTERGP 27 // L + letter group number
#define RULE_ALT1 28 // T word has $alt attribute
#define RULE_NOVOWELS 29 // X no vowels up to word boundary
#define RULE_SPELLING 31 // W while spelling letter-by-letter
#define RULE_LAST_RULE 31
#define LETTERGP_A 0
#define LETTERGP_B 1
#define LETTERGP_C 2
#define LETTERGP_H 3
#define LETTERGP_F 4
#define LETTERGP_G 5
#define LETTERGP_Y 6
#define LETTERGP_VOWEL2 7
// Punctuation types returned by ReadClause()
// bits 0-7 pause x 10mS, bits 8-10 intonation type,
// bit 15=sentence, bit 14=clause, bits 13=voice change
// bit 12 used to distinguish otherwise identical types
#define CLAUSE_BIT_SENTENCE 0x8000
#define CLAUSE_BIT_VOICE 0x2000
#define CLAUSE_NONE 0 + 0x0400
#define CLAUSE_PARAGRAPH 70 + 0x8000
#define CLAUSE_EOF 35 + 0x9000
#define CLAUSE_VOICE 0 + 0x2400
#define CLAUSE_PERIOD 35 + 0x8000
#define CLAUSE_COMMA 20 + 0x4100
#define CLAUSE_SHORTCOMMA 15 + 0x4100
#define CLAUSE_QUESTION 35 + 0x8200
#define CLAUSE_EXCLAMATION 40 + 0x8300
#define CLAUSE_COLON 30 + 0x4000
#ifdef PLATFORM_RISCOS
#define CLAUSE_SEMICOLON 30 + 0x4000
#else
#define CLAUSE_SEMICOLON 30 + 0x4100
#endif
#define SAYAS_CHARS 0x12
#define SAYAS_GLYPHS 0x13
#define SAYAS_SINGLE_CHARS 0x14
#define SAYAS_KEY 0x20
#define SAYAS_DIGITS1 0x21
#define SAYAS_DIGITS 0x30 // + number of digits
// Rule:
// [4] [match] [1 pre] [2 post] [3 phonemes] 0
// match 1 pre 2 post 0 - use common phoneme string
// match 1 pre 2 post 3 0 - empty phoneme string
typedef struct {
int points;
const char *phonemes;
int end_type;
char *del_fwd;
} MatchRecord;
// used to mark words with the source[] buffer
typedef struct{
unsigned short start;
unsigned short sourceix;
unsigned short flags;
unsigned char pre_pause;
unsigned char wmark;
unsigned char length;
} WORD_TAB;
// a clause translated into phoneme codes (first stage)
typedef struct {
unsigned char phcode;
unsigned char stress;
unsigned char tone_number;
unsigned char synthflags;
unsigned short sourceix;
} PHONEME_LIST2;
typedef struct {
int type;
int parameter[N_SPEECH_PARAM];
} PARAM_STACK;
extern PARAM_STACK param_stack[];
extern const int param_defaults[N_SPEECH_PARAM];
#define N_LOPTS 12
#define LOPT_DIERESES 1
// 1=remove [:] from unstressed syllables, 2= remove from unstressed or non-penultimate syllables
// bit 4=0, if stress < 4, bit 4=1, if not the highest stress in the word
#define LOPT_IT_LENGTHEN 2
// 1=german
#define LOPT_PREFIXES 3
// non-zero, change voiced/unoiced to match last consonant in a cluster
// bit 1=LANG=ru, don't propagate over [v]
// bit 2=don't propagate acress word boundaries
// bit 3=LANG=pl, propagate over liquids and nasals
#define LOPT_REGRESSIVE_VOICING 4
// 0=default, 1=no check, other allow this character as an extra initial letter (default is 's')
#define LOPT_UNPRONOUNCABLE 5
// select length_mods tables, (length_mod_tab) + (length_mod_tab0 * 100)
#define LOPT_LENGTH_MODS 6
// increase this to prevent sonorants being shortened before shortened (eg. unstressed) vowels
#define LOPT_SONORANT_MIN 7
// Italian "syntactic doubling"
#define LOPT_IT_DOUBLING 8
// max. amplitude for vowel at the end of a clause
#define LOPT_MAXAMP_EOC 9
// don't reduce the strongest vowel in a word which is marked 'unstressed'
#define LOPT_KEEP_UNSTR_VOWEL 10
// LANG=cs,sk combine some prepositions with the following word, if the combination has N or fewer syllables
// bits 0-3 N syllables
// bit 4=only if the second word has $alt attribute
// bit 5=not if the second word is end-of-sentence
#define LOPT_COMBINE_WORDS 11
typedef struct {
// bit0=don't link consonants with the next word
// bit1=don't add linking phonemes
// bit2=length of a final vowel doesn't depend on the next phoneme
// bit4=longer pause before STOP, VSTOP,FRIC
// bit8-11=separate words with adjacent stressed syllables with (1=pause, 2=long pausem 3=[?] phoneme)
int word_gap;
int vowel_pause;
int stress_rule; // 1=first syllable, 2=penultimate, 3=last
// bit0=don't stress monosyllables,
// bit1=don't set diminished stress,
// bit2=mark unstressed final syllables as diminished
// bit4=don't allow secondary stress on last syllable
// bit5-don't use automatic secondary stress
// bit6=light syllable followed by heavy, move secondary stress to the heavy syllable. LANG=Finnish
// bit8=stress last syllable if it doesn't end in a vowel
// bit9=stress last syllable if it doesn't end in vowel or "s" or "n" LANG=Spanish
// bit12= In a 2-syllable word, if one has primary stress then give the other secondary stress
// bit13= If there is only one syllable before the primary stress, give it a secondary stress
// bit16= Don't diminish consecutive syllables within a word.
int stress_flags;
int unstressed_wd1; // stress for $u word of 1 syllable
int unstressed_wd2; // stress for $u word of >1 syllable
int param[N_LOPTS];
unsigned char *length_mods;
unsigned char *length_mods0;
// bits0-1=which numbers routine to use.
// bit2= thousands separator must be space
// bit3= , decimal separator, not .
// bit4=use three-and-twenty rather than twenty-three
// bit5='and' between tens and units
// bit6=add "and" after hundred and thousand
// bit7=don't have "and" both after hundreds and also between tens and units
// bit8=only one primary stress in tens+units
// bit9=only one vowel betwen tens and units
// bit10=omit "one" before "hundred"
// bit11=don't say 19** as nineteen hundred
// bit12=allow space as thousands separator (in addition to langopts.thousands_sep)
// bits13-15 post-decimal-digits 0=single digits, 1=(LANG=it) 2=(LANG=pl) 3=(LANG=ro)
// bit16=dot after number indicates ordinal
// bit18=special word for 100,000s LANG=sw
// bit19=(LANG=pl)
int numbers;
// bits 1-4 use variant form of numbers before thousands,millions,etc.
int numbers2;
int thousands_sep;
int decimal_sep;
int intonation; // 1=tone language
int long_stop; // extra mS pause for a lengthened stop
int phoneme_change; // TEST, change phonemes, after translation
char max_initial_consonants;
char spelling_stress; // 0=default, 1=stress first letter
char tone_numbers;
char ideographs; // treat as separate words
int testing; // testing options: bit 1= specify stressed syllable in the form: "outdoor/2"
const wchar_t *replace_chars; // characters to be substitutes
const unsigned int *replacement_chars; // substitutes for replace_chars
} LANGUAGE_OPTIONS;
// a parameter of ChangePhonemes()
typedef struct {
int flags;
unsigned char stress; // stress level of this vowel
unsigned char stress_highest; // the highest stress level of a vowel in this word
unsigned char n_vowels; // number of vowels in the word
unsigned char vowel_this; // syllable number of this vowel (counting from 1)
unsigned char vowel_stressed; // syllable number of the highest stressed vowel
} CHANGEPH;
#define NUM_SEP_DOT 0x0008 // . , for thousands and decimal separator
#define NUM_SEP_SPACE 0x1000 // allow space as thousands separator (in addition to langopts.thousands_sep)
#define NUM_DEC_IT 0x2000 // (LANG=it) speak post-decimal-point digits as a combined number not as single digits
class Translator
{//=============
public:
Translator();
virtual ~Translator();
void *TranslateClause(FILE *f_text, const void *vp_input, int *tone, char **voice_change);
int TranslateWord(char *word, int next_pause, WORD_TAB *wtab);
int LoadDictionary(const char *name, int no_error);
virtual void CalcLengths();
virtual void CalcPitches(int clause_tone);
LANGUAGE_OPTIONS langopts;
int translator_name;
int transpose_offset;
int transpose_max;
int transpose_min;
char phon_out[300];
char word_phonemes[N_WORD_PHONEMES]; // a word translated into phoneme codes
int stress_amps[8];
int stress_amps_r[8];
int stress_lengths[8];
int dict_condition; // conditional apply some pronunciation rules and dict.lookups
const unsigned short *charset_a0; // unicodes for characters 0xa0 to oxff
const wchar_t *char_plus_apostrophe; // single chars + apostrophe treated as words
// holds properties of characters: vowel, consonant, etc for pronunciation rules
unsigned char letter_bits[256];
int letter_bits_offset;
#define N_LETTER_TYPES 20
const wchar_t *letter_groups[N_LETTER_TYPES];
/* index1=option, index2 by 0=. 1=, 2=?, 3=! 4=none */
unsigned char punct_to_tone[4][5];
char *data_dictrules; // language_1 translation rules file
char *data_dictlist; // language_2 dictionary lookup file
char *dict_hashtab[N_HASH_DICT]; // hash table to index dictionary lookup file
private:
int TranslateWord2(char *word, WORD_TAB *wtab, int pre_pause, int next_pause);
int TranslateLetter(char *letter, char *phonemes, int control);
void SetSpellingStress(char *phonemes, int control);
void GetTranslatedPhonemeString(char *phon_out, int n_phon_out);
void WriteMnemonic(int *ix, int mnem);
void MakePhonemeList(int post_pause, int new_sentence);
int SubstitutePhonemes(PHONEME_LIST2 *plist_out);
int ReadClause(FILE *f_in, char *buf, unsigned short *charix, int n_buf);
int AnnouncePunctuation(int c1, int c2, char *buf, int ix);
int LookupDict2(char *word, char *word2, char *phonetic, unsigned int *flags, int end_flags);
const char *LookupSpecial(char *string);
const char *LookupCharName(int c);
int LookupNum2(int value, int control, char *ph_out);
int LookupNum3(int value, char *ph_out, int suppress_null, int thousandplex, int prev_thousands);
int LookupThousands(int value, int thousandplex, char *ph_out);
int TranslateNumber_1(char *word1, char *ph_out, unsigned int *flags, int wflags);
void InitGroups(void);
void AppendPhonemes(char *string, int size, const char *ph);
char *DecodeRule(const char *group, char *rule);
void MatchRule(char *word[], const char *group, char *rule, MatchRecord *match_out, int end_flags, int dict_flags);
int TranslateRules(char *p, char *phonemes, int size, char *end_phonemes, int end_flags, int dict_flags);
void ApplySpecialAttribute(char *phonemes, int dict_flags);
int IsLetter(int letter, int group);
void CalcPitches_Tone(int clause_tone);
protected:
virtual int Unpronouncable(char *word);
virtual void SetWordStress(char *output, unsigned int dictionary_flags, int tonic, int prev_stress);
virtual int RemoveEnding(char *word, int end_type, char *word_copy);
virtual int TranslateChar(char *ptr, int prev_in, int c, int next_in, int *insert);
virtual int TranslateNumber(char *word1, char *ph_out, unsigned int *flags, int wflags);
virtual int ChangePhonemes(PHONEME_LIST2 *phlist, int n_ph, int index, PHONEME_TAB *ph, CHANGEPH *ch);
int IsVowel(int letter);
int LookupDictList(char *word1, char *ph_out, unsigned int *flags, int end_flags);
int Lookup(char *word, char *ph_out);
// groups1 and groups2 are indexes into data_dictrules, set up by InitGroups()
// the two-letter rules for each letter must be consecutive in the language_rules source
char *groups1[256]; // translation rule lists, index by single letter
char *groups2[N_RULE_GROUP2]; // translation rule lists, indexed by two-letter pairs
unsigned int groups2_name[N_RULE_GROUP2]; // the two letter pairs for groups2[]
int n_groups2; // number of groups2[] entries used
unsigned char groups2_count[256]; // number of 2 letter groups for this initial letter
unsigned char groups2_start[256]; // index into groups2
int n_ph_list2;
PHONEME_LIST2 ph_list2[N_PHONEME_LIST]; // first stage of text->phonemes
int expect_verb;
int expect_past; // expect past tense
int expect_verb_s;
int word_flags; // word is all upper case
int prev_last_stress;
int prepause_timeout;
int end_stressed_vowel; // word ends with stressed vowel
int prev_dict_flags; // dictionary flags from previous word
char *clause_end;
int word_vowel_count; // number of vowels so far
int word_stressed_count; // number of vowels so far which could be stressed
int clause_upper_count; // number of upper case letters in the clause
int clause_lower_count; // number of lower case letters in the clause
}; // end of class Translator
extern int option_tone1;
extern int option_tone2;
extern int option_waveout;
extern int option_quiet;
extern int option_phonemes;
extern int option_linelength; // treat lines shorter than this as end-of-clause
extern int option_harmonic1;
extern int option_multibyte;
extern int option_capitals;
extern int option_punctuation;
extern int option_endpause;
extern int option_ssml;
extern int option_phoneme_input; // allow [[phonemes]] in input text
extern int option_phoneme_variants;
extern int option_sayas;
extern int count_characters;
extern int count_words;
extern int count_sentences;
extern int skip_characters;
extern int skip_words;
extern int skip_sentences;
extern int skipping_text;
extern int end_character_position;
extern int clause_start_char;
extern int clause_start_word;
extern char *namedata;
#define N_MARKER_LENGTH 50 // max.length of a mark name
extern char skip_marker[N_MARKER_LENGTH];
#define N_PUNCTLIST 60
extern wchar_t option_punctlist[N_PUNCTLIST];
extern int speech_parameters[];
extern Translator *translator;
extern const unsigned short *charsets[N_CHARSETS];
extern char dictionary_name[40];
extern char ctrl_embedded; // to allow an alternative CTRL for embedded commands
extern char *p_textinput;
extern wchar_t *p_wchar_input;
extern int ungot_char;
extern int (* uri_callback)(int, const char *, const char *);
extern void SetLengthMods(Translator *tr, int value);
Translator *SelectTranslator(const char *name);
int CompileDictionary(const char *dsource, const char *dict_name, FILE *log, char *err_name);
void LoadConfig(void);
int PhonemeCode(unsigned int mnem);
void ChangeWordStress(char *word, int new_stress);
int TransposeAlphabet(char *text, int offset, int min, int max);
int utf8_in(int *c, char *buf, int backwards);
int utf8_out(unsigned int c, char *buf);
int lookupwchar(const unsigned short *list,int c);
int Eof(void);
char *strchr_w(const char *s, int c);
int IsBracket(int c);
void SetLetterBits(Translator *tr, int group, const char *string);
void SetLetterBitsRange(Translator *tr, int group, int first, int last);
void SetLetterVowel(Translator *tr, int c);
void InitNamedata(void);
void InitText(int flags);
void InitText2(void);
int IsDigit(unsigned int c);
int IsAlpha(unsigned int c);
int isspace2(unsigned int c);
extern FILE *f_trans; // for logging
Reference in New Issue View Git Blame Copy Permalink