void ProsodyDisplay::DrawEnv(wxDC& dc, int x1, int y1, int width, PHONEME_LIST *ph)
{//==============================================================================
int pitchr;
int pitch;
int p1;
int ix;
int x,y;
int y2=0;
unsigned char *env;
PHONEME_DATA phdata_tone;
if(width <= 0) return;
if((pitchr = ph->pitch2 - ph->pitch1) < 0)
{
pitchr = -pitchr;
p1 = ph->pitch2;
}
else
{
p1 = ph->pitch1;
}
if(p1 == 255) return;
dc.SetPen(PEN_PITCHENV);
env = envelope_data[ph->env];
if((ph->type == phVOWEL) && (ph->tone_ph != 0))
{
// the envelope is given by a Tone phoneme acting on this vowel
InterpretPhoneme2(ph->tone_ph, &phdata_tone);
env = GetEnvelope(phdata_tone.pitch_env);
}
if(env == NULL)
return;
for(ix=0; ix<=width; ix+=4)
{
x = int((ix * 127.9)/width);
pitch = p1 + (pitchr * env[x])/256;
y = y1-int(pitch * scaley);
if(ix > 0)
dc.DrawLine(x1+ix-4,y2,x1+ix,y);
y2 = y;
}
} // end of DrawEnv
void CalcLengths(Translator *tr)
{//==============================
int ix;
int ix2;
PHONEME_LIST *prev;
PHONEME_LIST *next;
PHONEME_LIST *next2;
PHONEME_LIST *next3;
PHONEME_LIST *p;
PHONEME_LIST *p2;
int stress;
int type;
static int more_syllables=0;
int pre_sonorant=0;
int pre_voiced=0;
int last_pitch = 0;
int pitch_start;
int length_mod;
int len;
int env2;
int end_of_clause;
int embedded_ix = 0;
int min_drop;
int pitch1;
int emphasized;
int tone_mod;
unsigned char *pitch_env=NULL;
PHONEME_DATA phdata_tone;
for(ix=1; ix<n_phoneme_list; ix++)
{
prev = &phoneme_list[ix-1];
p = &phoneme_list[ix];
stress = p->stresslevel & 0x7;
emphasized = p->stresslevel & 0x8;
next = &phoneme_list[ix+1];
if(p->synthflags & SFLAG_EMBEDDED)
{
DoEmbedded2(&embedded_ix);
}
type = p->type;
if(p->synthflags & SFLAG_SYLLABLE)
type = phVOWEL;
switch(type)
{
case phPAUSE:
last_pitch = 0;
break;
case phSTOP:
last_pitch = 0;
if(prev->type == phFRICATIVE)
p->prepause = 25;
else
if((more_syllables > 0) || (stress < 4))
p->prepause = 48;
else
p->prepause = 60;
if(prev->type == phSTOP)
p->prepause = 60;
if((tr->langopts.word_gap & 0x10) && (p->newword))
p->prepause = 60;
if(p->ph->phflags & phLENGTHENSTOP)
p->prepause += 30;
if(p->synthflags & SFLAG_LENGTHEN)
p->prepause += tr->langopts.long_stop;
break;
case phVFRICATIVE:
case phFRICATIVE:
if(p->newword)
{
if((prev->type == phVOWEL) && (p->ph->phflags & phNOPAUSE))
{
}
else
{
p->prepause = 15;
}
}
if(next->type==phPAUSE && prev->type==phNASAL && !(p->ph->phflags&phFORTIS))
p->prepause = 25;
if(prev->ph->phflags & phBRKAFTER)
p->prepause = 30;
if((tr->langopts.word_gap & 0x10) && (p->newword))
p->prepause = 30;
if((p->ph->phflags & phSIBILANT) && next->type==phSTOP && !next->newword)
{
if(prev->type == phVOWEL)
p->length = 200; // ?? should do this if it's from a prefix
else
p->length = 150;
}
else
p->length = 256;
if(type == phVFRICATIVE)
{
if(next->type==phVOWEL)
{
pre_voiced = 1;
}
if((prev->type==phVOWEL) || (prev->type == phLIQUID))
{
p->length = (255 + prev->length)/2;
}
}
break;
case phVSTOP:
if(prev->type==phVFRICATIVE || prev->type==phFRICATIVE || (prev->ph->phflags & phSIBILANT) || (prev->type == phLIQUID))
p->prepause = 30;
if(next->type==phVOWEL || next->type==phLIQUID)
{
if((next->type==phVOWEL) || !next->newword)
pre_voiced = 1;
p->prepause = 40;
if((prev->type == phPAUSE) || (prev->type == phVOWEL)) // || (prev->ph->mnemonic == ('/'*256+'r')))
p->prepause = 0;
else
if(p->newword==0)
{
if(prev->type==phLIQUID)
p->prepause = 20;
if(prev->type==phNASAL)
p->prepause = 12;
if(prev->type==phSTOP && !(prev->ph->phflags & phFORTIS))
p->prepause = 0;
}
}
if((tr->langopts.word_gap & 0x10) && (p->newword) && (p->prepause < 20))
p->prepause = 20;
break;
case phLIQUID:
case phNASAL:
p->amp = tr->stress_amps[0]; // unless changed later
p->length = 256; // TEMPORARY
min_drop = 0;
if(p->newword)
{
if(prev->type==phLIQUID)
p->prepause = 25;
if(prev->type==phVOWEL)
{
if(!(p->ph->phflags & phNOPAUSE))
p->prepause = 12;
}
}
if(next->type==phVOWEL)
{
pre_sonorant = 1;
}
else
{
p->pitch2 = last_pitch;
if((prev->type==phVOWEL) || (prev->type == phLIQUID))
{
p->length = prev->length;
if(p->type == phLIQUID)
{
p->length = speed1;
}
if(next->type == phVSTOP)
{
p->length = (p->length * 160)/100;
}
if(next->type == phVFRICATIVE)
{
p->length = (p->length * 120)/100;
}
}
else
{
for(ix2=ix; ix2<n_phoneme_list; ix2++)
{
if(phoneme_list[ix2].type == phVOWEL)
{
p->pitch2 = phoneme_list[ix2].pitch2;
break;
}
}
}
p->pitch1 = p->pitch2-16;
if(p->pitch2 < 16)
{
p->pitch1 = 0;
}
p->env = PITCHfall;
pre_voiced = 0;
}
break;
case phVOWEL:
min_drop = 0;
next2 = &phoneme_list[ix+2];
next3 = &phoneme_list[ix+3];
if(stress > 7) stress = 7;
if(stress <= 1)
{
stress = stress ^ 1; // swap diminished and unstressed (until we swap stress_amps,stress_lengths in tr_languages)
}
if(pre_sonorant)
p->amp = tr->stress_amps[stress]-1;
else
p->amp = tr->stress_amps[stress];
if(emphasized)
p->amp = 25;
if(ix >= (n_phoneme_list-3))
{
// last phoneme of a clause, limit its amplitude
if(p->amp > tr->langopts.param[LOPT_MAXAMP_EOC])
p->amp = tr->langopts.param[LOPT_MAXAMP_EOC];
}
// is the last syllable of a word ?
more_syllables=0;
end_of_clause = 0;
for(p2 = p+1; p2->newword== 0; p2++)
{
if((p2->type == phVOWEL) && !(p2->ph->phflags & phNONSYLLABIC))
more_syllables++;
if(p2->ph->code == phonPAUSE_CLAUSE)
end_of_clause = 2;
}
if(p2->ph->code == phonPAUSE_CLAUSE)
end_of_clause = 2;
if((p2->newword & 2) && (more_syllables==0))
{
end_of_clause = 2;
}
// calc length modifier
if((next->ph->code == phonPAUSE_VSHORT) && (next2->type == phPAUSE))
{
// if PAUSE_VSHORT is followed by a pause, then use that
next = next2;
next2 = next3;
next3 = &phoneme_list[ix+4];
}
if(more_syllables==0)
{
len = tr->langopts.length_mods0[next2->ph->length_mod *10+ next->ph->length_mod];
if((next->newword) && (tr->langopts.word_gap & 0x20))
{
// consider as a pause + first phoneme of the next word
length_mod = (len + tr->langopts.length_mods0[next->ph->length_mod *10+ 1])/2;
}
else
length_mod = len;
}
else
{
length_mod = tr->langopts.length_mods[next2->ph->length_mod *10+ next->ph->length_mod];
if((next->type == phNASAL) && (next2->type == phSTOP || next2->type == phVSTOP) && (next3->ph->phflags & phFORTIS))
length_mod -= 15;
}
if(more_syllables==0)
length_mod *= speed1;
else
if(more_syllables==1)
length_mod *= speed2;
else
length_mod *= speed3;
length_mod = length_mod / 128;
if(length_mod < 8)
length_mod = 8; // restrict how much lengths can be reduced
if(stress >= 7)
{
// tonic syllable, include a constant component so it doesn't decrease directly with speed
length_mod += tr->langopts.lengthen_tonic;
if(emphasized)
length_mod += (tr->langopts.lengthen_tonic/2);
}
else
if(emphasized)
{
length_mod += tr->langopts.lengthen_tonic;
}
if((len = tr->stress_lengths[stress]) == 0)
len = tr->stress_lengths[6];
length_mod = length_mod * len;
if(p->tone_ph != 0)
{
if((tone_mod = phoneme_tab[p->tone_ph]->std_length) > 0)
{
// a tone phoneme specifies a percentage change to the length
length_mod = (length_mod * tone_mod) / 100;
}
}
if((end_of_clause == 2) && !(tr->langopts.stress_flags & S_NO_EOC_LENGTHEN))
{
// this is the last syllable in the clause, lengthen it - more for short vowels
len = (p->ph->std_length * 2);
if(tr->langopts.stress_flags & 0x40000)
len=200; // don't lengthen short vowels more than long vowels at end-of-clause
length_mod = length_mod * (256 + (280 - len)/3)/256;
}
if(length_mod > tr->langopts.max_lengthmod*speed1)
{
//limit the vowel length adjustment for some languages
length_mod = (tr->langopts.max_lengthmod*speed1);
}
length_mod = length_mod / 128;
if(p->type != phVOWEL)
{
length_mod = 256; // syllabic consonant
min_drop = 16;
}
p->length = length_mod;
if(p->env >= (N_ENVELOPE_DATA-1))
{
fprintf(stderr,"espeak: Bad intonation data\n");
p->env = 0;
}
// pre-vocalic part
// set last-pitch
env2 = p->env + 1; // version for use with preceding semi-vowel
if(p->tone_ph != 0)
{
InterpretPhoneme2(p->tone_ph, &phdata_tone);
pitch_env = GetEnvelope(phdata_tone.pitch_env);
}
else
{
pitch_env = envelope_data[env2];
}
pitch_start = p->pitch1 + ((p->pitch2-p->pitch1)*pitch_env[0])/256;
if(pre_sonorant || pre_voiced)
{
// set pitch for pre-vocalic part
if(pitch_start == 255)
last_pitch = pitch_start; // pitch is not set
if(pitch_start - last_pitch > 16)
last_pitch = pitch_start - 16;
prev->pitch1 = last_pitch;
prev->pitch2 = pitch_start;
if(last_pitch < pitch_start)
{
prev->env = PITCHrise;
p->env = env2;
}
else
{
prev->env = PITCHfall;
}
prev->length = length_mod;
prev->amp = p->amp;
if((prev->type != phLIQUID) && (prev->amp > 18))
prev->amp = 18;
}
// vowel & post-vocalic part
next->synthflags &= ~SFLAG_SEQCONTINUE;
if(next->type == phNASAL && next2->type != phVOWEL)
next->synthflags |= SFLAG_SEQCONTINUE;
if(next->type == phLIQUID)
{
next->synthflags |= SFLAG_SEQCONTINUE;
if(next2->type == phVOWEL)
{
next->synthflags &= ~SFLAG_SEQCONTINUE;
}
if(next2->type != phVOWEL)
{
if(next->ph->mnemonic == ('/'*256+'r'))
{
next->synthflags &= ~SFLAG_SEQCONTINUE;
// min_drop = 15;
}
}
}
if((min_drop > 0) && ((p->pitch2 - p->pitch1) < min_drop))
{
pitch1 = p->pitch2 - min_drop;
if(pitch1 < 0)
pitch1 = 0;
p->pitch1 = pitch1;
}
last_pitch = p->pitch1 + ((p->pitch2-p->pitch1)*envelope_data[p->env][127])/256;
pre_sonorant = 0;
pre_voiced = 0;
break;
}
}
} // end of CalcLengths
