protected override void BeforeConsumption()
{
Log(LogLevel.MajorInfo, "Started");
base.BeforeConsumption();
curElementIndex = -1;
curSegment = curWord = null;
curSyl = null;
LoadStressHeuristics(stressHeuristicsFile);
}
protected override void BeforeConsumption()
{
Console.WriteLine("segmenter started");
base.BeforeConsumption();
curElementIndex = -1;
curSegment = curWord = null;
curSyl = null;
LoadStressHeuristics(stressHeuristicsFile);
}
public Word(List <SpeechElement> Elements)
{
Phonemes = Elements;
List <SpeechElement> elems = new List <SpeechElement>(Phonemes);
while ((elems.Count > 0) && (elems[elems.Count - 1] is Separator))
{
elems.RemoveAt(elems.Count - 1);
}
List <SpeechElement> nuclei = elems.FindAll(
delegate(SpeechElement e)
{
return((e is Vowel) && (!((Vowel)e).IsVowelIn(Vowels.VeryShort | Vowels.Inaudible | Vowels.PatahGnuva)));
}
);
Syllable syl;
if (nuclei.Count > 0)
{
foreach (SpeechElement nucleus in nuclei)
{
int i = elems.IndexOf(nucleus);
syl = new Syllable(this, (Vowel)nucleus, i, i);
for (int j = i + 1; j < elems.Count; j++)
{
if (elems[j] is Consonant)
{
if (((((Consonant)elems[j]).Flags & ConsonantFlags.StrongDagesh) != 0) ||
(j + 1 == elems.Count) ||
((elems[j + 1] is Vowel) &&
((Vowel)elems[j + 1]).IsVowelIn(Vowels.Inaudible | Vowels.PatahGnuva)))
{
syl.HintStrongDagesh = (((Consonant)elems[j]).Flags & ConsonantFlags.StrongDagesh) != 0;
syl.End = j;
}
else
{
break;
}
}
else if ((elems[j] is Vowel) && (((Vowel)elems[j]).IsVowelIn(Vowels.Inaudible | Vowels.PatahGnuva)))
{
syl.End = j;
}
else
{
break;
}
}
Syllables.Add(syl);
}
syl = null;
for (int i = 0; i < Syllables.Count; i++)
{
syl = Syllables[i];
if (i == 0)
{
syl.Start = 0;
}
else if (Syllables[i - 1].HintStrongDagesh)
{
syl.Start = Syllables[i - 1].End;
}
else
{
syl.Start = Syllables[i - 1].End + 1;
}
}
if (syl != null)
{
if (syl.End < elems.Count - 1)
{
syl.End = elems.Count - 1;
}
}
List <SpeechElement> cant = elems.FindAll(delegate(SpeechElement e)
{
return(e is Cantillation);
}
);
foreach (Cantillation c in cant)
{
int ci = Phonemes.IndexOf(c);
int si = Syllables.FindLastIndex(delegate(Syllable s)
{
return(s.Start < ci);
}
);
Syllables[si].CantillationMarks.Add(c.Mark);
this.CantillationMarks.Add(c.Mark);
Phonemes.RemoveAt(ci);
Syllables.ForEach(delegate(Syllable s)
{
if (s.Start > ci)
{
s.Start--;
}
if (s.End >= ci)
{
s.End--;
}
});
}
}
}
public bool Match(Syllable syl)
{
return(syl.Match(this));
}
protected override void Consume(Queue <Segment> InQueue)
{
if (InQueue.Count == 0)
{
return;
}
Segment seg = InQueue.Dequeue();
_ItemConsumed(seg);
Segment nextSeg = null;
if (InQueue.Count > 0)
{
nextSeg = InQueue.Peek();
}
if (seg is SeparatorSegment)
{
if (HebrewChar.IsPunctuation((seg[0] as Separator).Latin[0]))
{
Phone phn = Phone.Create(seg[0]);
if (phn == null)
{
phn = new Phone("_", 1);
}
Emit(phn);
}
firstStressInClause = true;
}
else if (seg is Word)
{
/*bool nextIsPunctuation = false;
* if (nextSeg is SeparatorSegment)
* nextIsPunctuation = HebrewChar.IsPunctuation(nextSeg[0].Latin[0]);*/
Word w = (Word)seg;
bool beforeStress = true;
SpeechElement HintStrongDagesh = null;
for (int sylIndex = 0; sylIndex < w.Syllables.Count; sylIndex++)
{
Syllable syl = w.Syllables[sylIndex];
bool stressed = syl.IsStressed;
if (beforeStress && stressed)
{
beforeStress = false;
}
bool beforeNucleus = true;
//bool heavySyl = ((syl.Coda == SyllableCoda.Closed) ^ ((syl.Nucleus!=null) && syl.Nucleus.IsVowelIn(Vowels.Long)));
// bool sylStart=true;
//foreach (SpeechElement e in w.Phonemes.GetRange(syl.Start,syl.End-syl.Start+1)) {
for (int elemIndex = 0; elemIndex < syl.Phonemes.Count; elemIndex++)
{
SpeechElement e = syl.Phonemes[elemIndex];
Phone phone = null;
// bool sylEnd = (elemIndex == syl.Phonemes.Count - 1);
if ((e is Consonant) && ((((Consonant)e).Flags & ConsonantFlags.StrongDagesh) != 0) && (e == HintStrongDagesh))
{
continue;
}
phone = Phone.Create(e);
if (phone == null)
{
continue;
}
phone.Context.SylIndex = sylIndex;
phone.Context.SylReverseIndex = w.Syllables.Count - sylIndex - 1;
phone.Context.IsNucleus = (syl.Nucleus == e);
phone.Context.IsAccented = stressed;
if (nextSeg is SeparatorSegment)
{
phone.Context.NextSeparator = (nextSeg as SeparatorSegment)[0].Latin;
}
if (!stressed)
{
if (Options.Akanye)
{
if (phone.Symbol == "o")
{
phone.Symbol = "a";
}
}
if (Options.Ikanye)
{
if (phone.Symbol == "e")
{
phone.Symbol = "i";
}
}
}
else
{
phone.Context.AccentStrength = 1;
}
if (phone.PitchCurve.Count > 0)
{
phone.PitchCurve.Clear();
}
phone.Duration = 80;
if (e is Vowel)
{
Vowel v = (Vowel)e;
if (v.IsVowelIn(Vowels.VeryShort))
{
if (v.vowel == Vowels.AudibleSchwa)
{
phone.Duration = 28;
}
else
{
phone.Duration = 40;
}
}
else if (v.IsVowelIn(Vowels.Short))
{
phone.Duration = 90;
}
else if (v.IsVowelIn(Vowels.Long))
{
phone.Duration = 94;
}
else if (v.IsVowelIn(Vowels.VeryLong))
{
phone.Duration = 97;
}
if (v.IsVowelIn(Vowels.HighVowels))
{
phone.Duration += 25;
}
else
{
phone.Duration += 30;
}
}
else if (e is Consonant)
{
phone.Duration = (((Consonant)e).Sonority) * 1.6 + 60;
}
else
{
phone.Duration = 100;
}
if (stressed)
{
if (firstStressInClause)
{
firstStressInClause = false;
}
if (e == syl.Nucleus)
{
beforeNucleus = false;
}
phone.Duration *= 1;
if (beforeNucleus && (e is Consonant) && ((Consonant)e).IsLiquid)
{
phone.Duration *= 1.2;
}
}
else
{
if (e is Vowel)
{
if (beforeStress)
{
phone.Duration *= 0.85;
}
else
{
phone.Duration *= 0.9;
}
}
}
if ((e is Consonant) && ((((Consonant)e).Flags & ConsonantFlags.StrongDagesh) != 0))
{
HintStrongDagesh = e;
if (Options.DistinguishStrongDagesh)
{
phone.Duration *= 1.4;
}
else
{
phone.Duration *= 1.1F;
}
}
/* if ((!nextIsPunctuation) && !heavySyl)
* phone.Duration *= 0.7;*/
// if ((nextSeg is SeparatorSegment) && (sylIndex == w.Syllables.Count - 1))
// phone.Duration *= 1.6;
Emit(phone);
// sylStart=false;
}
}
/* if (w.CantillationMarks.Contains('֑'))
* silpr.Phones.Add(new Phone("_",200));*/
#region cantillation stuff
if (Options.SingCantillation)
{
foreach (char ch in w.CantillationMarks)
{
int i = HebrewChar.DisjunctiveRank(ch), len = 0;
if (i < 5)
{
switch (i)
{
case 1:
len = 230;
break;
case 2:
len = 165;
break;
case 3:
len = 60;
break;
case 4:
len = 30;
break;
}
Emit(new Phone("_", len));
}
}
}
#endregion
}
}
protected override void Consume(Queue <SpeechElement> InQueue)
{
if (InQueue.Count == 0)
{
return;
}
SpeechElement curElement = InQueue.Dequeue();
Log("Consuming {0} ({1})", curElement.GetType().Name, curElement.Latin);
_ItemConsumed(curElement);
SpeechElement nextElement = null;
if (InQueue.Count > 0)
{
nextElement = InQueue.Peek();
}
if (curSegment == null)
{
if ((curElement is Phoneme) || (curElement is WordTag))
{
curSegment = new Word();
}
else
{
curSegment = new SeparatorSegment();
}
curElementIndex = -1;
}
if (curSegment is Word)
{
curWord = (curSegment as Word);
if (curElement is Phoneme)
{
curWord.Add(curElement);
curElementIndex++;
if (curSyl == null)
{
curSyl = new Syllable(curWord);
curSyl.Start = curElementIndex;
curSyl.End = curElementIndex;
}
else if (curSyl.Nucleus == null)
{
curSyl.End++;
if ((curElement is Vowel) &&
(curElement as Vowel).IsVowelIn(Vowels.LegalNuclei))
{
curSyl.Nucleus = curElement as Vowel;
}
}
else if (curElement is Consonant)
{
if ((nextElement != null) && (nextElement is Vowel) && (nextElement as Vowel).IsVowelIn(Vowels.Inaudible))
{
curSyl.End++;
nextElement = InQueue.Dequeue();
Log("Assimilating a {1}", nextElement.GetType().Name, (nextElement as Vowel).vowel);
}
else if ((nextElement == null) || (nextElement is Separator))
{
curSyl.End++;
// curWord.Syllables.Add(curSyl);
// curSyl=null;
}
else
{
if (((curElement as Consonant).Flags & ConsonantFlags.StrongDagesh) != 0)
{
curSyl.End++;
curSyl.HintStrongDagesh = true;
}
curWord.Syllables.Add(curSyl);
curSyl = new Syllable(curWord);
curSyl.Start = curElementIndex;
curSyl.End = curElementIndex;
}
}
else
{
// non-nucleic vowel, move on
curSyl.End++;
}
}
else if (curElement is WordTag)
{
curWord.Tag = (curElement as WordTag).Tag;
}
else if (curElement is Qaryan.Core.Cantillation)
{
curWord.CantillationMarks.Add((curElement as Cantillation).Mark);
}
else if (curElement is Separator)
{
AddAndProcessWord(curSegment as Word);
curSegment = null;
}
}
if (curElement is Separator)
{
if (curSegment != null)
{
if (curSegment is SeparatorSegment)
{
curSegment.Add(curElement);
}
else
{
if (curSegment is Word)
{
AddAndProcessWord(curSegment as Word);
}
else
{
Emit(curSegment);
}
curSegment = new SeparatorSegment(curElement as Separator);
}
Log("Adding separator segment");
Emit(curSegment);
curSegment = null;
}
else
{
Log("Adding separator segment");
Emit(curSegment = new SeparatorSegment(curElement as Separator));
curSegment = null;
}
}
}
/// <summary>
/// Performs stress assignment and related processing on a word, and adds it to the queue of produced segments.
/// </summary>
/// <param name="w">The word obtained from the segmentation step.</param>
void AddAndProcessWord(Word w)
{
if ((curSyl != null) && !w.Syllables.Contains(curSyl))
{
w.Syllables.Add(curSyl);
curSyl = null;
}
SyllablePattern?sp = w.PlaceStress(StressHeuristics, DefaultStress);
// Log(LogLevel.Debug, sp);
bool beforeStress = true;
foreach (Syllable syl in w.Syllables)
{
bool stressed = syl.IsStressed;
if (stressed)
{
beforeStress = false;
}
for (int i = 0; i < syl.Phonemes.Count; i++)
{
SpeechElement e = syl.Phonemes[i];
if (e is Vowel)
{
Vowel v = (Vowel)e;
if (v.vowel == Vowels.KamatzIndeterminate)
{
if (beforeStress && (syl.Coda == SyllableCoda.Closed) && (syl.Phonemes[syl.Phonemes.Count - 1] is Consonant) /*&& ((w.Tag&TagTypes.Origin)!=TagTypes.Foreign)*/ && !stressed)
{
v.vowel = Vowels.KamatzKatan;
}
else
{
v.vowel = Vowels.KamatzGadol;
}
}
else if (v.vowel == Vowels.AudibleSchwa)
{
if ((w.Tag & TagTypes.Origin) == TagTypes.Foreign)
{
v.vowel = Vowels.SilentSchwa;
}
else
{
int j = w.Phonemes.IndexOf(e);
if ((j + 1 < w.Phonemes.Count) &&
(w.Phonemes[j + 1] is Consonant) &&
(j - 1 >= 0) &&
(w.Phonemes[j - 1] is Consonant) &&
(w.Phonemes[j + 1].Latin != w.Phonemes[j - 1].Latin))
{
switch (w.Phonemes[j - 1].Latin)
{
/* case "k":
* case "l":
* case "b":
* case "m":
* break;*/
case Consonants.Vav:
case Consonants.Lamed:
break;
default:
if (!relaxAudibleSchwa)
{
break;
}
int son = ((Consonant)w.Phonemes[j + 1]).Sonority -
((Consonant)w.Phonemes[j - 1]).Sonority;
if (son >= 0)
{
v.Silent = true;
}
break;
}
}
}
}
}
}
}
if (sp != null)
{
Log(LogLevel.Info, "/{0}/ from pattern {1}", w.TranslitSyllablesStress, sp.ToString());
}
else
{
Log(LogLevel.Info, "/{0}/", w.TranslitSyllablesStress);
}
Emit(w);
}
void AddAndProcessWord(Word w)
{
if ((curSyl != null) && !w.Syllables.Contains(curSyl))
{
w.Syllables.Add(curSyl);
curSyl = null;
}
Console.WriteLine("segmenter: /{0}/", w.TranslitSyllables);
w.PlaceStress(StressHeuristics, DefaultStress);
foreach (Syllable syl in w.Syllables)
{
bool stressed = syl.IsStressed;
for (int i = 0; i < syl.Phonemes.Count; i++)
{
SpeechElement e = syl.Phonemes[i];
if (e is Vowel)
{
Vowel v = (Vowel)e;
if (v.vowel == Vowels.KamatzIndeterminate)
{
if ((syl.Coda == SyllableCoda.Closed) && (syl.Phonemes[syl.Phonemes.Count - 1] is Consonant) /*&& ((w.Tag&TagTypes.Origin)!=TagTypes.Foreign)*/ && !stressed)
{
v.vowel = Vowels.KamatzKatan;
}
else
{
v.vowel = Vowels.KamatzGadol;
}
// Log.Analyzer.WriteLine("Kamatz determined to be "+v.vowel.ToString());
}
else if (v.vowel == Vowels.AudibleSchwa)
{
if ((w.Tag & TagTypes.Origin) == TagTypes.Foreign)
{
v.vowel = Vowels.SilentSchwa;
}
else
{
int j = w.Phonemes.IndexOf(e);
if ((j + 1 < w.Phonemes.Count) &&
(w.Phonemes[j + 1] is Consonant) &&
(j - 1 >= 0) &&
(w.Phonemes[j - 1] is Consonant) &&
(w.Phonemes[j + 1].Latin != w.Phonemes[j - 1].Latin))
{
switch (w.Phonemes[j - 1].Latin)
{
/* case "k":
* case "l":
* case "b":
* case "m":
* break;*/
case Consonants.Vav:
case Consonants.Lamed:
break;
default:
if (!relaxAudibleSchwa)
{
break;
}
int son = ((Consonant)w.Phonemes[j + 1]).Sonority -
((Consonant)w.Phonemes[j - 1]).Sonority;
if (son >= 0)
{
v.Silent = true;
}
break;
}
}
}
}
}
}
}
Emit(w);
}
