mjohnson/espeak-ng
1
0
Fork 0
mirror of https://github.com/rhasspy/espeak-ng.git synced 2026-11-10 08:13:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'development'

Browse Source 
Conflicts:
	phsource/phonemes
This commit is contained in:
Reece H. Dunn
2014-03-04 17:51:51 +00:00
parent 38f02aa611 7592c683a8
commit b10823e1bc
46 changed files with 2386 additions and 1011 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/compiledata.cpp
+4 -1
View File
@@ -294,7 +294,7 @@ static keywtab_t keywords[] = {
{"Continue", tSTATEMENT, kCONTINUE},
{"ChangePhoneme", tINSTRN1, i_CHANGE_PHONEME},
{"ReplaceNextPhoneme", tINSTRN1, i_REPLACE_NEXT_PHONEME},
{"ChangeNextPhoneme", tINSTRN1, i_REPLACE_NEXT_PHONEME},
{"InsertPhoneme", tINSTRN1, i_INSERT_PHONEME},
{"AppendPhoneme", tINSTRN1, i_APPEND_PHONEME},
{"IfNextVowelAppend", tINSTRN1, i_APPEND_IFNEXTVOWEL},
@@ -1133,6 +1133,9 @@ static unsigned int StringToWord(const char *string)
unsigned char c;
unsigned int word;
if(string==NULL)
return(0);
word = 0;
for(ix=0; ix<4; ix++)
{
src/compiledict.cpp
+20 -6
View File
@@ -50,8 +50,9 @@ static char *hash_chains[N_HASH_DICT];
static char letterGroupsDefined[N_LETTER_GROUPS];
MNEM_TAB mnem_rules[] = {
{"unpr", 0x01},
{"noprefix", 0x02}, // rule fails if a prefix has been removed
{"unpr", DOLLAR_UNPR},
{"noprefix", DOLLAR_NOPREFIX}, // rule fails if a prefix has been removed
{"list", DOLLAR_LIST}, // a pronunciation is given in the *_list file
{"w_alt1", 0x11},
{"w_alt2", 0x12},
@@ -314,6 +315,7 @@ char *DecodeRule(const char *group_chars, int group_length, char *rule, int cont
value = *rule++ & 0xff;
if((value != 0x01) || (control & FLAG_UNPRON_TEST))
{
// TODO write the string backwards if in RULE_PRE
p[0] = '$';
name = LookupMnemName(mnem_rules, value);
strcpy(&p[1],name);
@@ -1102,21 +1104,33 @@ static void copy_rule_string(char *string, int *state_out)
break;
case '$':
output[ix++] = RULE_DOLLAR;
c = 0;
value = 0;
mr = mnem_rules;
while(mr->mnem != NULL)
{
len = strlen(mr->mnem);
if(memcmp(p, mr->mnem, len) == 0)
{
c = mr->value;
value = mr->value;
p += len;
break;
}
mr++;
}
if(c == 0)
if(state == 1)
{
// pre-rule, put the number before the RULE_DOLLAR
output[ix++] = value;
c = RULE_DOLLAR;
}
else
{
output[ix++] = RULE_DOLLAR;
c = value;
}
if(value == 0)
{
fprintf(f_log,"%5d: $ command not recognized\n",linenum);
error_count++;
src/dictionary.cpp
+36 -8
View File
@@ -1929,6 +1929,7 @@ static void MatchRule(Translator *tr, char *word[], char *word_start, int group_
int add_points;
int command;
int check_atstart;
unsigned int *flags;
MatchRecord match;
static MatchRecord best;
@@ -2166,11 +2167,11 @@ static void MatchRule(Translator *tr, char *word[], char *word_start, int group_
case RULE_DOLLAR:
command = *rule++;
if(command == 0x01)
if(command == DOLLAR_UNPR)
{
match.end_type = SUFX_UNPRON; // $unpron
}
else if(command == 0x02) // $noprefix
else if(command == DOLLAR_NOPREFIX) // $noprefix
{
if(word_flags & FLAG_PREFIX_REMOVED)
failed = 1; // a prefix has been removed
@@ -2185,16 +2186,20 @@ static void MatchRule(Translator *tr, char *word[], char *word_start, int group_
else
failed = 1;
}
else if((command & 0xf0) == 0x20)
else if(((command & 0xf0) == 0x20) || (command == DOLLAR_LIST))
{
// $p_alt
// $list or $p_alt
// make a copy of the word up to the post-match characters
ix = *word - word_start + consumed + group_length + 1;
memcpy(word_buf, word_start-1, ix);
word_buf[ix] = ' ';
word_buf[ix+1] = 0;
LookupFlags(tr, &word_buf[1], &flags);
if(LookupFlags(tr, &word_buf[1]) & (1 << (BITNUM_FLAG_ALT + (command & 0xf))))
if((command == DOLLAR_LIST) && (flags[0] & FLAG_FOUND) && !(flags[1] & FLAG_ONLY))
add_points = 23;
else
if(flags[0] & (1 << (BITNUM_FLAG_ALT + (command & 0xf))))
add_points = 23;
else
failed = 1;
@@ -2408,6 +2413,28 @@ static void MatchRule(Translator *tr, char *word[], char *word_start, int group_
failed = 1;
break;
case RULE_DOLLAR:
command = *rule++;
if((command==DOLLAR_LIST) || ((command & 0xf0) == 0x20))
{
// $list or $p_alt
// make a copy of the word up to the current character
ix = *word - word_start + 1;
memcpy(word_buf, word_start-1, ix);
word_buf[ix] = ' ';
word_buf[ix+1] = 0;
LookupFlags(tr, &word_buf[1], &flags);
if((command==DOLLAR_LIST) && (flags[0] & FLAG_FOUND) && !(flags[1] & FLAG_ONLY))
add_points = 23;
else
if(flags[0] & (1 << (BITNUM_FLAG_ALT + (command & 0xf))))
add_points = 23;
else
failed = 1;
}
break;
case RULE_SYLLABLE:
/* more than specified number of vowels to the left */
syllable_count = 1;
@@ -3586,14 +3613,15 @@ int Lookup(Translator *tr, const char *word, char *ph_out)
}
int LookupFlags(Translator *tr, const char *word)
{//==============================================
int LookupFlags(Translator *tr, const char *word, unsigned int **flags_out)
{//===========================================================================
char buf[100];
static unsigned int flags[2];
char *word1 = (char *)word;
flags[0] = flags[1] = 0;
LookupDictList(tr, &word1, buf, flags, 0, NULL);
*flags_out = flags;
return(flags[0]);
}
@@ -3654,7 +3682,7 @@ int RemoveEnding(Translator *tr, char *word, int end_type, char *word_copy)
}
// remove bytes from the end of the word and replace them by spaces
for(i=0; (i<len_ending) && (i < sizeof(ending)-1); i++)
for(i=0; (i<len_ending) && (i < (int)sizeof(ending)-1); i++)
{
ending[i] = word_end[i];
word_end[i] = ' ';
src/espeakedit.cpp
+7 -1
View File
@@ -1,5 +1,5 @@
/***************************************************************************
* Copyright (C) 2005 to 2013 by Jonathan Duddington *
* Copyright (C) 2005 to 2014 by Jonathan Duddington *
* email: jonsd@users.sourceforge.net *
* Copyright (C) 2013 by Reece H. Dunn *
* *
@@ -139,6 +139,12 @@ if(argc > 1)
if(strcmp(param,"--compile")==0)
{
samplerate_native = samplerate = 22050;
LoadPhData(NULL);
if(LoadVoice("", 0) == NULL)
{
fprintf(stderr, "Failed to load default voice\n");
exit(1);
}
CompilePhonemeData();
CompileIntonation();
}
src/phonemelist.cpp
+8
View File
@@ -399,6 +399,14 @@ void MakePhonemeList(Translator *tr, int post_pause, int start_sentence)
InterpretPhoneme(tr, 0x100, plist3, &phdata, &worddata);
if((alternative = phdata.pd_param[pd_CHANGE_NEXTPHONEME]) > 0)
{
ph_list3[j+1].ph = phoneme_tab[alternative];
ph_list3[j+1].phcode = alternative;
ph_list3[j+1].type = phoneme_tab[alternative]->type;
next = phoneme_tab[alternative];
}
if(((alternative = phdata.pd_param[pd_INSERTPHONEME]) > 0) && (inserted == 0))
{
// PROBLEM: if we insert a phoneme before a vowel then we loose the stress.
src/prosodydisplay.cpp
+13 -10
View File
@@ -1,5 +1,5 @@
/***************************************************************************
* Copyright (C) 2005 to 2013 by Jonathan Duddington *
* Copyright (C) 2005 to 2014 by Jonathan Duddington *
* email: jonsd@users.sourceforge.net *
* *
* This program is free software; you can redistribute it and/or modify *
@@ -343,19 +343,22 @@ void ProsodyDisplay::OnMouse(wxMouseEvent& event)
line = pt.y / FRAMEHEIGHT;
// find which phoneme is selected on this line
for(ix=linetab[line]; (ix<linetab[line+1]) && (ix<numph); ix++)
if(line < num_lines)
{
xpos += GetWidth(&phlist[ix]);
if(xpos > pt.x)
// find which phoneme is selected on this line
for(ix=linetab[line]; (ix<linetab[line+1]) && (ix<numph); ix++)
{
selected_ph = ix;
SelectPh(selected_ph);
break;
xpos += GetWidth(&phlist[ix]);
if(xpos > pt.x)
{
selected_ph = ix;
SelectPh(selected_ph);
break;
}
}
}
RefreshLine(line);
RefreshLine(line);
}
if(event.RightDown())
{
src/readclause.cpp
+15 -3
View File
@@ -1,5 +1,5 @@
/***************************************************************************
* Copyright (C) 2005 to 2013 by Jonathan Duddington *
* Copyright (C) 2005 to 2014 by Jonathan Duddington *
* email: jonsd@users.sourceforge.net *
* *
* This program is free software; you can redistribute it and/or modify *
@@ -2092,7 +2092,13 @@ static int ProcessSsmlTag(wchar_t *xml_buf, char *outbuf, int *outix, int n_outb
}
if((attr2 = GetSsmlAttribute(px,"time")) != NULL)
{
value = (attrnumber(attr2,0,1) * 25) / speed.pause_factor; // compensate for speaking speed to keep constant pause length
value2 = attrnumber(attr2,0,1); // pause in mS
// compensate for speaking speed to keep constant pause length, see function PauseLength()
// 'value' here is x 10mS
value = (value2 * 256) / (speed.clause_pause_factor * 10);
if(value < 200)
value = (value2 * 256) / (speed.pause_factor * 10);
if(terminator == 0)
terminator = CLAUSE_NONE;
@@ -2100,7 +2106,13 @@ static int ProcessSsmlTag(wchar_t *xml_buf, char *outbuf, int *outix, int n_outb
if(terminator)
{
if(value > 0xfff)
value = 0xfff;
{
// scale down the value and set a scaling indicator bit
value = value / 32;
if(value > 0xfff)
value = 0xfff;
terminator |= CLAUSE_PAUSE_LONG;
}
return(terminator + value);
}
break;
src/synthdata.cpp
+1 -1
View File
@@ -35,7 +35,7 @@
#include "translate.h"
#include "wave.h"
const char *version_string = "1.48.02 02.Feb.14";
const char *version_string = "1.48.03 04.Mar.14";
const int version_phdata = 0x014801;
int option_device_number = -1;
src/synthesize.cpp
+14 -4
View File
@@ -1,5 +1,5 @@
/***************************************************************************
* Copyright (C) 2005 to 2013 by Jonathan Duddington *
* Copyright (C) 2005 to 2014 by Jonathan Duddington *
* email: jonsd@users.sourceforge.net *
* *
* This program is free software; you can redistribute it and/or modify *
@@ -188,7 +188,7 @@ static void DoPitch(unsigned char *env, int pitch1, int pitch2)
int PauseLength(int pause, int control)
{//====================================
int len;
unsigned int len;
if(control == 0)
{
@@ -210,8 +210,10 @@ int PauseLength(int pause, int control)
static void DoPause(int length, int control)
{//=========================================
// length in nominal mS
// control = 1, less shortening at fast speeds
int len;
unsigned int len;
int srate2;
if(length == 0)
len = 0;
@@ -219,7 +221,15 @@ static void DoPause(int length, int control)
{
len = PauseLength(length, control);
len = (len * samplerate) / 1000; // convert from mS to number of samples
if(len < 90000)
{
len = (len * samplerate) / 1000; // convert from mS to number of samples
}
else
{
srate2 = samplerate / 25; // avoid overflow
len = (len * srate2) / 40;
}
}
EndPitch(1);
src/synthesize.h
+4 -3
View File
@@ -1,5 +1,5 @@
/***************************************************************************
* Copyright (C) 2005 to 2013 by Jonathan Duddington *
* Copyright (C) 2005 to 2014 by Jonathan Duddington *
* email: jonsd@users.sourceforge.net *
* *
* This program is free software; you can redistribute it and/or modify *
@@ -212,7 +212,7 @@ typedef struct {
unsigned char tone_ph; // tone phoneme to use with this vowel
PHONEME_TAB *ph;
short length; // length_mod
unsigned int length; // length_mod
unsigned char env; // pitch envelope number
unsigned char type;
unsigned char prepause;
@@ -239,6 +239,7 @@ typedef struct {
#define pd_INSERTPHONEME i_INSERT_PHONEME
#define pd_APPENDPHONEME i_APPEND_PHONEME
#define pd_CHANGEPHONEME i_CHANGE_PHONEME
#define pd_CHANGE_NEXTPHONEME i_REPLACE_NEXT_PHONEME
#define pd_LENGTHMOD i_SET_LENGTH
#define pd_FORNEXTPH 0x2
@@ -379,7 +380,7 @@ typedef struct {
typedef struct {
int pause_factor;
int clause_pause_factor;
int min_pause;
unsigned int min_pause;
int wav_factor;
int lenmod_factor;
int lenmod2_factor;
src/tr_languages.cpp
+1 -1
View File
@@ -781,7 +781,7 @@ Translator *SelectTranslator(const char *name)
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 ??
tr->langopts.numbers = NUM_SINGLE_STRESS | NUM_DECIMAL_COMMA | NUM_AND_UNITS | NUM_HUNDRED_AND | NUM_OMIT_1_HUNDRED | NUM_VIGESIMAL;
tr->langopts.numbers = NUM_SINGLE_STRESS | NUM_DECIMAL_COMMA | NUM_HUNDRED_AND | NUM_OMIT_1_HUNDRED | NUM_OMIT_1_THOUSAND | NUM_VIGESIMAL;
}
break;
src/translate.cpp
+3 -1
View File
@@ -104,7 +104,6 @@ static int embedded_read;
unsigned int embedded_list[N_EMBEDDED_LIST];
// the source text of a single clause (UTF8 bytes)
#define N_TR_SOURCE 800
static char source[N_TR_SOURCE+40]; // extra space for embedded command & voice change info at end
int n_replace_phonemes;
@@ -2655,6 +2654,9 @@ void *TranslateClause(Translator *tr, FILE *f_text, const void *vp_input, int *t
charix[charix_top+3] = 0;
clause_pause = (terminator & 0xfff) * 10; // mS
if(terminator & CLAUSE_PAUSE_LONG)
clause_pause = clause_pause * 32 ; // pause value is *320mS not *10mS
tone = (terminator >> 12) & 0x7;
if(tone2 != 0)
{
src/translate.h
+13 -2
View File
@@ -26,6 +26,9 @@
#define N_WORD_PHONEMES 200 // max phonemes in a word
#define N_WORD_BYTES 160 // max bytes for the UTF8 characters in a word
#define N_CLAUSE_WORDS 300 // max words in a clause
#define N_TR_SOURCE 800 // the source text of a single clause (UTF8 bytes)
#define N_RULE_GROUP2 120 // max num of two-letter rule chains
#define N_HASH_DICT 1024
#define N_CHARSETS 20
@@ -180,6 +183,11 @@
#define RULE_SPELLING 31 // W while spelling letter-by-letter
#define RULE_LAST_RULE 31
#define DOLLAR_UNPR 0x01
#define DOLLAR_NOPREFIX 0x02
#define DOLLAR_LIST 0x03
#define LETTERGP_A 0
#define LETTERGP_B 1
#define LETTERGP_C 2
@@ -191,7 +199,7 @@
// Punctuation types returned by ReadClause()
// bits 0-7 pause x 10mS, bits 12-14 intonation type,
// bits 0-11 pause x 10mS
// bits12-14 intonation type
// bit 15- don't need space after the punctuation
// bit 19=sentence, bit 18=clause, bits 17=voice change
@@ -199,6 +207,8 @@
// bit 20= punctuation character can be inside a word (Armenian)
// bit 21= speak the name of the punctuation character
// bit 22= dot after the last word
// bit 23= pause is x 320mS (not x 10mS)
#define CLAUSE_BIT_SENTENCE 0x80000
#define CLAUSE_BIT_CLAUSE 0x40000
#define CLAUSE_BIT_VOICE 0x20000
@@ -206,6 +216,7 @@
#define PUNCT_IN_WORD 0x100000
#define PUNCT_SAY_NAME 0x200000
#define CLAUSE_DOT 0x400000
#define CLAUSE_PAUSE_LONG 0x800000
#define CLAUSE_NONE ( 0 + 0x04000)
#define CLAUSE_PARAGRAPH (70 + 0x80000)
@@ -741,7 +752,7 @@ int SetTranslator2(const char *name);
void DeleteTranslator(Translator *tr);
void ProcessLanguageOptions(LANGUAGE_OPTIONS *langopts);
int Lookup(Translator *tr, const char *word, char *ph_out);
int LookupFlags(Translator *tr, const char *word);
int LookupFlags(Translator *tr, const char *word, unsigned int **flags_out);
int TranslateNumber(Translator *tr, char *word1, char *ph_out, unsigned int *flags, WORD_TAB *wtab, int control);
int TranslateRoman(Translator *tr, char *word, char *ph_out, WORD_TAB *wtab);
src/voices.cpp
+1 -1
View File
@@ -1,5 +1,5 @@
/***************************************************************************
* Copyright (C) 2005 to 2013 by Jonathan Duddington *
* Copyright (C) 2005 to 2014 by Jonathan Duddington *
* email: jonsd@users.sourceforge.net *
* *
* This program is free software; you can redistribute it and/or modify *