JNIEXPORT jobjectArray JNICALL Java_ee_filosoft_vabamorf_Linguistic_synthesize(JNIEnv *env, jobject jobj, jobject info, jstring hint)
{
FUNCTION_HEADER;
if (!linguistic) return NULL;
try {
jclass MorphInfoClass = env->FindClass("ee/filosoft/vabamorf/Linguistic$MorphInfo");
jmethodID MorphInfoClassConstructor = (MorphInfoClass ? env->GetMethodID(MorphInfoClass, "<init>", "()V") : 0);
if (!MorphInfoClassConstructor) return NULL;
CMorphInfo morphInfo;
morphInfo.m_szRoot = FSJNIStrtoW(env, FSJNIGetStringField(env, info, "root"));
morphInfo.m_szClitic = FSJNIStrtoW(env, FSJNIGetStringField(env, info, "clitic"));
morphInfo.m_cPOS = FSJNIGetCharField(env, info, "pos");
morphInfo.m_szForm = FSJNIStrtoW(env, FSJNIGetStringField(env, info, "form"));
CFSArray<CMorphInfo> results = linguistic->Synthesize(morphInfo, FSJNIStrtoW(env, hint));
jobjectArray synths=(jobjectArray)env->NewObjectArray(results.GetSize(), env->FindClass("java/lang/Object"), 0);
for (INTPTR ip=0; ip<results.GetSize(); ip++) {
jobject analyze1=env->NewObject(MorphInfoClass, MorphInfoClassConstructor);
FSJNISetStringField(env, analyze1, "root", FSJNIWtoStr(env, results[ip].m_szRoot));
FSJNISetStringField(env, analyze1, "ending", FSJNIWtoStr(env, results[ip].m_szEnding));
FSJNISetStringField(env, analyze1, "clitic", FSJNIWtoStr(env, results[ip].m_szClitic));
FSJNISetCharField(env, analyze1, "pos", results[ip].m_cPOS);
FSJNISetStringField(env, analyze1, "form", FSJNIWtoStr(env, results[ip].m_szForm));
env->SetObjectArrayElement(synths, ip, analyze1);
}
return synths;
} catch(...) {
return NULL;
}
}
// add morphological analysis to CFSArray containing the sentence
void addAnalysis(CLinguistic& linguistic, CDisambiguator& disambiguator, CFSArray<CFSVar>& words, const bool disambiguate) {
//CFSVar &words=Data["words"];
CFSArray<CPTWord> PTWords;
for (INTPTR ip=0; ip<words.GetSize(); ip++) {
PTWords.AddItem(words[ip]["text"].GetWString());
}
// perform analysis and optional disambiguation
CFSArray<CMorphInfos> MorphResults=linguistic.AnalyzeSentense(PTWords);
if (disambiguate) {
MorphResults=disambiguator.Disambiguate(MorphResults);
}
// collect the analysis results
ASSERT(PTWords.GetSize()==MorphResults.GetSize());
for (INTPTR ip=0; ip<words.GetSize(); ip++) {
const CFSArray<CMorphInfo> &Analysis=MorphResults[ip].m_MorphInfo;
CFSVar VarAnalysis;
VarAnalysis.Cast(CFSVar::VAR_ARRAY);
for (INTPTR ipRes=0; ipRes<Analysis.GetSize(); ipRes++) {
const CMorphInfo &Analysis1=Analysis[ipRes];
CFSVar VarAnalysis1;
VarAnalysis1["root"]=Analysis1.m_szRoot;
VarAnalysis1["ending"]=Analysis1.m_szEnding;
VarAnalysis1["clitic"]=Analysis1.m_szClitic;
VarAnalysis1["partofspeech"]=CFSWString(Analysis1.m_cPOS);
VarAnalysis1["form"]=Analysis1.m_szForm;
VarAnalysis[ipRes]=VarAnalysis1;
}
words[ip]["analysis"]=VarAnalysis;
}
}
void do_phones(syl_struct &ss) {
CFSArray<phone_struct> pv;
phone_struct p;
p.utt_p = 0;
p.phr_p = 0;
p.word_p = 0;
p.syl_p = 0;
INTPTR phone_syl_p = 1;
for (INTPTR i = 0; i < ss.syl.GetLength(); i++) {
CFSWString c = ss.syl.GetAt(i);
if ((c == L'š') || (c == L'ž')) c = L"sh";
else
if (c == L'q') c = L"kw";
else
if (c == L'õ') c = L"q";
else
if (c == L'ä') c = L"x";
else
if (c == L'ö') c = L"c";
else
if (c == L'ü') c = L"y";
if (is_colon(c)) {
if (i > 0) // nihutusvigade kaitseks (vt "piirkonda")
pv[pv.GetSize() - 1].phone += doq;
} else {
p.phone = c;
p.syl_p = phone_syl_p++;
pv.AddItem(p);
}
}
ss.phone_vector = pv;
}
void do_phones(syl_struct &ss) {
CFSArray<phone_struct> pv;
phone_struct p;
p.utt_p = 0;
p.phr_p = 0;
p.word_p = 0;
p.syl_p = 0;
INTPTR phone_syl_p = 1;
for (INTPTR i = 0; i < ss.syl.GetLength(); i++) {
CFSWString c = ss.syl.GetAt(i);
if ((c == L'š') || (c == L'ž')) c = L"sh";
if (c == L'õ') c = L"q";
if (c == L'ä') c = L"x";
if (c == L'ö') c = L"c";
if (c == L'ü') c = L"y";
if (is_colon(c)) {
// 1. nihutusvigade kaitseks (vt "piirkonda")
// 2-3. kolmandas vältes v ja j on kõnebaasis sedavõrd haruldased,
// et väljundis kuuleb nende asemel mingit r-i laadset hääikut.
// Kellel on parem baas, kommenteerigu 2. ja 3. tingimus välja.
if ((i > 0) && (pv[pv.GetSize() - 1].phone != L"j") && (pv[pv.GetSize() - 1].phone != L"v"))
pv[pv.GetSize() - 1].phone += doq;
} else {
p.phone = c;
p.syl_p = phone_syl_p++;
pv.AddItem(p);
}
}
ss.phone_vector = pv;
}
CFSArray<CMorphInfos> convertDisambInput(std::vector<WordAnalysis> const& words) {
CFSArray<CMorphInfos> infosarray;
const int n = words.size();
for (int i=0 ; i<n ; ++i) {
WordAnalysis const& word = words[i];
infosarray.AddItem(convertWordAnalysis(word));
}
return infosarray;
}
void phrase2words(phrase_struct &p, CFSArray<CPTWord> &PTW) {
CFSWString res;
CFSArray<CPTWord> PTWords;
CFSArray<CFSWString> word_array;
for (INTPTR i = 0; i < p.s.GetLength(); i++) {
CFSWString c = p.s.GetAt(i);
if (is_space(c) || i == (p.s.GetLength() - 1)) {
if (i == (p.s.GetLength() - 1)) res += c;
word_array.AddItem(res);
res = empty_str;
} else res += c;
}
word_array = tokens2words(word_array);
p.s = empty_str;
for (INTPTR i = 0; i < word_array.GetSize(); i++) {
p.s += word_array[i];
if (i < word_array.GetSize() - 1) p.s += sp;
PTW.AddItem(word_array[i]); //.ToLower()); Kui väiketähtedeks, siis siin
}
p.word_c = word_array.GetSize();
}
// convert output to wrapper format
std::vector<SpellingResults> convertSpellingOutput(CFSArray<CFSVar>& words) {
std::vector<SpellingResults> results;
results.reserve(words.GetSize());
for (int widx=0 ; widx < words.GetSize() ; ++widx) {
CFSVar word = words[widx];
std::string text = std::string(word["text"].GetAString());
CFSVar suggestions = word["suggestions"];
StringVector suggestStrings;
suggestStrings.reserve(suggestions.GetSize());
for (int sidx=0 ; sidx < suggestions.GetSize() ; ++sidx) {
CFSVar suggestion = suggestions[sidx];
suggestStrings.push_back(std::string(suggestion.GetAString()));
}
results.push_back(SpellingResults(text, word["spelling"].GetInt(), suggestStrings));
}
return results;
}
JNIEXPORT jobjectArray JNICALL Java_ee_filosoft_vabamorf_Linguistic_analyze(JNIEnv *env, jobject jobj, jstring word)
{
FUNCTION_HEADER;
if (!linguistic) return NULL;
try {
CFSArray<CMorphInfo> results = linguistic->Analyze(FSJNIStrtoW(env, word));
jobjectArray analyzes=(jobjectArray)env->NewObjectArray(results.GetSize(), env->FindClass("java/lang/Object"), 0);
for (INTPTR ip=0; ip<results.GetSize(); ip++) {
env->SetObjectArrayElement(analyzes, ip, MorphInfoToJNI(env, results[ip]));
}
return analyzes;
} catch(...) {
return NULL;
}
}
// convert vabamorf base library output to WordAnalysis instances, which as easier to wrap.
std::vector<WordAnalysis> convertOutput(CFSArray<CFSVar>& words) {
std::vector<WordAnalysis> results;
results.reserve(words.GetSize());
for (int widx=0 ; widx < words.GetSize() ; ++widx) {
CFSVar word = words[widx];
CFSVar analysis = word["analysis"];
AnalysisVector vec;
for (int aidx=0 ; aidx < analysis.GetSize() ; ++aidx) {
CFSVar a = analysis[aidx];
vec.push_back(Analysis(a["root"].GetAString(),
a["ending"].GetAString(),
a["clitic"].GetAString(),
a["partofspeech"].GetAString(),
a["form"].GetAString()));
}
results.push_back(WordAnalysis(std::string(word["text"].GetAString()), vec));
}
return results;
}
std::vector<WordAnalysis> convertDisambOutput(CFSArray<CMorphInfos> const& morphInfos) {
std::vector<WordAnalysis> output;
const int n = morphInfos.GetSize();
output.reserve(n);
for (int i=0 ; i<n ; ++i) {
CMorphInfos const& infos = morphInfos[i];
output.push_back(convertMorphInfos(infos));
}
return output;
}
CFSArray<CFSWString> tokens2words(CFSArray<CFSWString> a) {
CFSArray<CFSWString> temp_array;
CFSArray<CFSWString> res;
for (INTPTR i = 0; i < a.GetSize(); i++) {
CFSWString s = a[i];
if (is_abbreviation(s, temp_array) > 0) {
for (INTPTR i = 0; i < temp_array.GetSize(); i++)
res.AddItem(temp_array[i]);
} else
if (is_word(s))
res.AddItem(s);
else {
CFSArray<CFSWString> carray;
make_ctype_array(s, carray);
for (INTPTR ictype = 0; ictype < carray.GetSize(); ictype++) {
INTPTR c_type = ctype(carray[ictype].GetAt(0));
if (c_type == 1) { //Tähed
if (is_word(carray[ictype])) res.AddItem(carray[ictype]);
else { // kui on sodipodi
explode(carray[ictype], L"", temp_array);
for (INTPTR i_temp = 0; i_temp < temp_array.GetSize(); i_temp++)
res.AddItem(replace_schar(temp_array[i_temp]));
}
} else
if (c_type == 2) { //Sümbolid
explode(carray[ictype], L"", temp_array);
for (INTPTR i_temp = 0; i_temp < temp_array.GetSize(); i_temp++)
res.AddItem(replace_schar(temp_array[i_temp]));
} else
if (c_type == 3) { //Numbrid
CFSWString nr = int_to_words(carray[ictype]);
explode(nr, sp, temp_array);
for (INTPTR i_temp = 0; i_temp < temp_array.GetSize(); i_temp++)
if (temp_array[i_temp].GetLength() > 0)
res.AddItem(temp_array[i_temp]);
}
}
}
} // iga token
return res;
}
JNIEXPORT jobjectArray JNICALL Java_ee_filosoft_vabamorf_Linguistic_synthesize(JNIEnv *env, jobject jobj, jobject info, jstring hint)
{
FUNCTION_HEADER;
if (!linguistic) return NULL;
try {
CMorphInfo MorphInfo = JNIToMorphInfo(env, info);
MorphInfo.m_szEnding.Empty();
CFSArray<CMorphInfo> results = linguistic->Synthesize(MorphInfo, FSJNIStrtoW(env, hint));
jobjectArray synths=(jobjectArray)env->NewObjectArray(results.GetSize(), env->FindClass("java/lang/Object"), 0);
for (INTPTR ip=0; ip<results.GetSize(); ip++) {
env->SetObjectArrayElement(synths, ip, MorphInfoToJNI(env, results[ip]));
}
return synths;
} catch(...) {
return NULL;
}
}
CMorphInfo DisambiguateSynthesisResult(CFSArray<CMorphInfo> MIs, CPTWord s) {
CFSWString s1;
CMorphInfo x;
for (INTPTR i = 0; i < MIs.GetSize(); i++) {
s1 = MIs[i].m_szRoot + MIs[i].m_szEnding + MIs[i].m_szClitic;
if (s.m_szWord == make_char_string(s1)) {
MIs[i].m_szRoot = s1;
return MIs[i];
}
}
x.m_szRoot = s.m_szWord;
return x;
}
JNIEXPORT jobjectArray JNICALL Java_ee_filosoft_vabamorf_Linguistic_analyzeSentence(JNIEnv *env, jobject jobj, jobjectArray words)
{
FUNCTION_HEADER;
if (!linguistic) return NULL;
try {
CFSArray<CPTWord> ptwords;
INTPTR ipSize = env->GetArrayLength(words);
for (INTPTR ip=0; ip<ipSize; ip++) {
ptwords.AddItem(FSJNIStrtoW(env, (jstring)env->GetObjectArrayElement(words, ip)));
}
CFSArray<CMorphInfos> results = linguistic->AnalyzeSentence(ptwords);
jobjectArray analyzes = (jobjectArray)env->NewObjectArray(results.GetSize(), env->FindClass("java/lang/Object"), 0);
for (INTPTR ip = 0; ip < results.GetSize(); ip++) {
env->SetObjectArrayElement(analyzes, ip, MorphInfosToJNI(env, results[ip]));
}
return analyzes;
} catch(...) {
return NULL;
}
}
JNIEXPORT jintArray JNICALL Java_ee_filosoft_vabamorf_Linguistic_spellWords(JNIEnv *env, jobject jobj, jobjectArray words)
{
FUNCTION_HEADER;
if (!linguistic) return NULL;
try {
CFSArray<CPTWord> ptwords;
INTPTR ipSize = env->GetArrayLength(words);
for (INTPTR ip=0; ip<ipSize; ip++) {
ptwords.AddItem(FSJNIStrtoW(env, (jstring)env->GetObjectArrayElement(words, ip)));
}
CFSArray<SPLRESULT> splresults = linguistic->SpellWords(ptwords);
jintArray result = env->NewIntArray(splresults.GetSize());
for (INTPTR ip=0; ip<splresults.GetSize(); ip++) {
jint result1 = splresults[ip];
env->SetIntArrayRegion(result, ip, 1, &result1);
}
return result;
} catch (...) {
return NULL;
}
}
void add_stress2(CFSArray<syl_struct> &sv, INTPTR wp) {
/* Kõige radikaalsem rõhutus siiani.
* wp = kui on liitsõna esimene liige siis on seal pearõhk.
*/
INTPTR main_stress = 2;
INTPTR stress = 1;
INTPTR size = sv.GetSize();
INTPTR stress_type = extra_stress(sv, size);
if (stress_type == 0) {
for (INTPTR i = 0; i < size; i++) {
if (i % 2 == 0) {
if ((i == 0) && (wp == 0))
sv[i].stress = main_stress;
else
sv[i].stress = stress;
}
}
if (size > 1) sv[size - 1].stress = 0;
}
else {
if (wp == 0)
sv[stress_type].stress = main_stress;
else
sv[stress_type].stress = stress;
//esimene pool
if (stress_type == 1) sv[0].stress = 0;
else
for (INTPTR i = stress_type - 1; i >= 0; i--)
if (i % 2 == 0)
sv[i].stress = stress;
if ((stress_type % 2 != 0) && (stress_type > 1))
sv[stress_type - 1].stress = 0;
//teine pool
INTPTR lopp = size - stress_type;
if (lopp > 3) {
for (INTPTR i = stress_type + 1; i < size; i++)
if (i % 2 != 0) sv[i].stress = stress;
sv[size - 1].stress = 0;
}
}
}
// spellcheck the words and add suggestions
void addSuggestions(CLinguistic& linguistic, CFSArray<CFSVar>& words, const bool suggest) {
for (INTPTR ip=0; ip<words.GetSize(); ip++) {
CFSVar &Word=words[ip];
CPTWord PTWord=Word["text"].GetWString();
PTWord.RemoveHyphens();
PTWord.RemovePunctuation();
PTWord.Trim();
if (PTWord.m_szWord.IsEmpty() || linguistic.SpellWord(PTWord.m_szWord)==SPL_NOERROR) {
Word["spelling"]=true;
} else {
Word["spelling"]=false;
if (suggest) {
CFSWStringArray Suggestions=linguistic.Suggest(PTWord.m_szWord);
CFSVar VarSuggestions;
VarSuggestions.Cast(CFSVar::VAR_ARRAY);
for (INTPTR ipRes=0; ipRes<Suggestions.GetSize(); ipRes++) {
VarSuggestions[ipRes]=Suggestions[ipRes];
}
Word["suggestions"]=VarSuggestions;
}
}
}
}
void OnValReadEnd(const CFSAString &szKey, CFSVar &Data) {
if (szKey.IsEmpty()) {
SubKeys("paragraphs", Data);
m_Writer.ObjectEnd();
} else if (szKey=="/paragraphs") {
m_Writer.ArrayEnd();
m_iCollectData++;
} else if (KeyMatch(szKey, "/paragraphs/%d")) {
SubKeys("sentences", Data);
m_Writer.ObjectEnd();
m_iCollectData--;
} else if (KeyMatch(szKey, "/paragraphs/%d/sentences")) {
m_Writer.ArrayEnd();
m_iCollectData++;
} else if (KeyMatch(szKey, "/paragraphs/%d/sentences/%d")) {
if (Data.KeyExist("words")) {
CFSVar &Words=Data["words"];
CFSArray<CMorphInfos> WordsAnalysis;
for (INTPTR ip=0; ip<Words.GetSize(); ip++) {
const CFSVar &Word=Words[ip];
CMorphInfos Analysis;
Analysis.m_szWord=Word["text"].GetWString();
const CFSVar &VarAnalysis=Word["analysis"];
for (INTPTR ip2=0; ip2<VarAnalysis.GetSize(); ip2++) {
const CFSVar &VarAnalysis1=VarAnalysis[ip2];
CMorphInfo Analysis1;
Analysis1.m_szRoot=VarAnalysis1["root"].GetWString();
Analysis1.m_szEnding=VarAnalysis1["ending"].GetWString();
Analysis1.m_szClitic=VarAnalysis1["clitic"].GetWString();
Analysis1.m_cPOS=VarAnalysis1["partofspeech"].GetWString()[0];
Analysis1.m_szForm=VarAnalysis1["form"].GetWString();
Analysis.m_MorphInfo.AddItem(Analysis1);
}
WordsAnalysis.AddItem(Analysis);
}
WordsAnalysis=m_Disambiguator.Disambiguate(WordsAnalysis);
RT_ASSERT(Words.GetSize()==WordsAnalysis.GetSize());
for (INTPTR ip=0; ip<Words.GetSize(); ip++) {
const CMorphInfos &Analysis=WordsAnalysis[ip];
CFSVar VarAnalysis;
VarAnalysis.Cast(CFSVar::VAR_ARRAY);
for (INTPTR ipRes=0; ipRes<Analysis.m_MorphInfo.GetSize(); ipRes++) {
const CMorphInfo &Analysis1=Analysis.m_MorphInfo[ipRes];
CFSVar VarAnalysis1;
VarAnalysis1["root"]=Analysis1.m_szRoot;
VarAnalysis1["ending"]=Analysis1.m_szEnding;
VarAnalysis1["clitic"]=Analysis1.m_szClitic;
VarAnalysis1["partofspeech"]=CFSWString(Analysis1.m_cPOS);
VarAnalysis1["form"]=Analysis1.m_szForm;
VarAnalysis[ipRes]=VarAnalysis1;
}
Words[ip]["analysis"]=VarAnalysis;
}
}
m_Writer.Val(Data);
m_iCollectData--;
}
}
CFSArray<CFSWString> do_all(CFSWString utt, bool print_label, bool print_utt) {
CFSArray<CFSWString> res;
CFSArray<CPTWord> PTW;
utterance_struct u;
u.s = utt.ToLower();
u.syl_c = 0;
u.phone_c = 0;
u.phra_c = do_phrases(u);
INTPTR word_count = 0;
if (print_utt) fprintf(stderr, "%s\n", ccstr(utt));
for (INTPTR i = 0; i < u.phr_vector.GetSize(); i++) {
u.phr_vector[i].utt_p = i;
phrase2words(u.phr_vector[i], PTW);
word_count += u.phr_vector[i].word_c;
}
CFSArray<CMorphInfo> words;
for (INTPTR i = 0; i < PTW.GetSize(); i++) {
CMorphInfo MI;
MI.m_szRoot = PTW[i].m_szWord;
words.AddItem(MI);
}
u.word_c = words.GetSize();
word_struct w;
INTPTR utt_phone_c = 1;
INTPTR syl_utt_p = 1;
INTPTR phone_utt_p = 1;
for (INTPTR i = 0; i < u.phr_vector.GetSize(); i++) {
u.phr_vector[i].utt_p = i + 1;
INTPTR syl_phr_p = 1;
INTPTR phone_phr_p = 1;
INTPTR phrase_pho_c = 1;
for (INTPTR i1 = 0; i1 < u.phr_vector[i].word_c; i1++) {
w.utt_p = utt_phone_c++;
w.phr_p = i1 + 1;
w.mi = words[0];
w.mi.m_szRoot += make_char_string(w.mi.m_szEnding) + w.mi.m_szClitic;
w.mi.m_szRoot = w.mi.m_szRoot.ToLower();
do_syls(w);
u.phr_vector[i].word_vector.AddItem(w);
INTPTR phone_word_p = 1;
INTPTR word_pho_c = 1;
for (INTPTR i2 = 0; i2 < u.phr_vector[i].word_vector[i1].syl_vector.GetSize(); i2++) {
u.syl_c++;
u.phr_vector[i].syl_c++;
u.phr_vector[i].word_vector[i1].syl_c++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].phr_p = syl_phr_p++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].utt_p = syl_utt_p++;
INTPTR syl_phone_c = 1;
do_phones(u.phr_vector[i].word_vector[i1].syl_vector[i2]);
for (INTPTR i3 = 0; i3 < u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector.GetSize(); i3++) {
u.phone_c++;
u.phr_vector[i].phone_c = phrase_pho_c++;
u.phr_vector[i].word_vector[i1].phone_c = word_pho_c++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_c = syl_phone_c++;
phone_struct p = u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector[i3];
p.utt_p = phone_utt_p++;
p.phr_p = phone_phr_p++;
p.word_p = phone_word_p++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector[i3] = p;
}
}
words.RemoveItem(0, 1);
}
}
if (print_label) print_u(u);
res = do_label(u);
return res;
}
int main(int argc, char* argv[]) {
size_t num_voices;
char **fn_voices;
char* in_fname;
char* output_fname;
FILE * outfp;
char* dur_fname;
FILE * durfp;
bool print_label = false;
bool print_utt = false;
bool write_raw = false;
bool write_durlabel = false;
CFSAString LexFileName, LexDFileName;
HTS_Engine engine;
double speed = 1.1;
size_t fr = 48000;
size_t fp = 240;
float alpha = 0.55;
float beta = 0.0;
float ht = 2.0;
float th = 0.5;
float gvw1 = 1.0;
float gvw2 = 1.2;
FSCInit();
fn_voices = (char **) malloc(argc * sizeof (char *));
if (argc < 11) {
fprintf(stderr, "Viga: liiga vähe parameetreid\n\n");
PrintUsage();
}
for (int i = 0; i < argc; i++) {
if (CFSAString("-lex") == argv[i]) {
if (i + 1 < argc) {
LexFileName = argv[++i];
} else {
return PrintUsage();
}
}
if (CFSAString("-lexd") == argv[i]) {
if (i + 1 < argc) {
LexDFileName = argv[++i];
} else {
return PrintUsage();
}
}
if (CFSAString("-m") == argv[i]) {
if (i + 1 < argc) {
fn_voices[0] = argv[i + 1];
} else {
fprintf(stderr, "Viga: puudub *.htsvoice fail\n");
PrintUsage();
exit(0);
}
}
if (CFSAString("-o") == argv[i]) {
if (i + 1 < argc) {
output_fname = argv[i + 1];
cfileexists(output_fname);
} else {
fprintf(stderr, "Viga: puudb väljundfaili nimi\n");
PrintUsage();
exit(0);
}
}
if (CFSAString("-f") == argv[i]) {
if (i + 1 < argc) {
in_fname = argv[i + 1];
} else {
fprintf(stderr, "Viga: puudb sisendfaili nimi\n");
PrintUsage();
exit(0);
}
}
if (CFSAString("-s") == argv[i]) {
if (i + 1 < argc) {
samplerate(fr, fp, alpha, atoi(argv[i + 1]));
}
}
if (CFSAString("-r") == argv[i]) {
if (i + 1 < argc) {
speed = atof(argv[i + 1]);
}
}
if (CFSAString("-ht") == argv[i]) {
if (i + 1 < argc) {
ht = atof(argv[i + 1]);
}
}
if (CFSAString("-gvw1") == argv[i]) {
if (i + 1 < argc) {
gvw1 = atof(argv[i + 1]);
}
}
if (CFSAString("-gvw2") == argv[i]) {
if (i + 1 < argc) {
gvw2 = atof(argv[i + 1]);
}
}
if (CFSAString("-debug") == argv[i]) {
print_label = true;
}
if (CFSAString("-utt") == argv[i]) {
print_utt = true;
}
if (CFSAString("-raw") == argv[i]) {
write_raw = true;
}
if (CFSAString("-dur") == argv[i]) {
if (i + 1 < argc) {
dur_fname = argv[i + 1];
cfileexists(dur_fname);
write_durlabel = true;
} else {
fprintf(stderr, "Viga: puudb kestustefaili nimi\n");
PrintUsage();
exit(0);
}
}
}
Linguistic.Open(LexFileName);
Disambiguator.Open(LexDFileName);
CFSWString text;
ReadUTF8Text(text, in_fname);
HTS_Engine_initialize(&engine);
if (HTS_Engine_load(&engine, fn_voices, 1) != TRUE) {
fprintf(stderr, "Viga: puudub *.htsvoice. %p\n", fn_voices[0]);
free(fn_voices);
HTS_Engine_clear(&engine);
exit(1);
}
free(fn_voices);
HTS_Engine_set_sampling_frequency(&engine, (size_t) fr);
HTS_Engine_set_phoneme_alignment_flag(&engine, FALSE);
HTS_Engine_set_fperiod(&engine, (size_t) fp);
HTS_Engine_set_alpha(&engine, alpha);
HTS_Engine_set_beta(&engine, beta);
HTS_Engine_set_speed(&engine, speed);
HTS_Engine_add_half_tone(&engine, ht);
HTS_Engine_set_msd_threshold(&engine, 1, th);
/*
HTS_Engine_set_duration_interpolation_weight(&engine, 1, diw);
HTS_Engine_set_parameter_interpolation_weight(&engine, 0, 0, piw1);
HTS_Engine_set_parameter_interpolation_weight(&engine, 0, 1, piw2);
HTS_Engine_set_gv_interpolation_weight(&engine, 0, 0, giw1);
HTS_Engine_set_gv_interpolation_weight(&engine, 0, 1, giw2);
*/
HTS_Engine_set_gv_weight(&engine, 0, gvw1);
HTS_Engine_set_gv_weight(&engine, 1, gvw2);
text = DealWithText(text);
CFSArray<CFSWString> res = do_utterances(text);
INTPTR data_size = 0;
outfp = fopen(output_fname, "wb");
if (write_durlabel) durfp = fopen(dur_fname, "w");
if (!write_raw) HTS_Engine_write_header(&engine, outfp, 1);
for (INTPTR i = 0; i < res.GetSize(); i++) {
CFSArray<CFSWString> label = do_all(res[i], print_label, print_utt);
std::vector<std::string> v;
v = to_vector(label);
std::vector<char*> vc;
fill_char_vector(v, vc);
size_t n_lines = vc.size();
if (HTS_Engine_synthesize_from_strings(&engine, &vc[0], n_lines) != TRUE) {
fprintf(stderr, "Viga: süntees ebaonnestus.\n");
HTS_Engine_clear(&engine);
exit(1);
}
clean_char_vector(vc);
data_size += HTS_Engine_engine_speech_size(&engine);
if (write_durlabel) HTS_Engine_save_durlabel(&engine, durfp);
HTS_Engine_save_generated_speech(&engine, outfp);
HTS_Engine_refresh(&engine);
} //synth loop
if (!write_raw) HTS_Engine_write_header(&engine, outfp, data_size);
if (write_durlabel) fclose(durfp);
fclose(outfp);
HTS_Engine_clear(&engine);
Linguistic.Close();
FSCTerminate();
return 0;
}
std::vector<std::string> to_vector(CFSArray<CFSWString> arr) {
std::vector<std::string> v;
for (INTPTR i = 0; i < arr.GetSize(); i++)
v.push_back(to_stdstring(arr[i]));
return v;
}
CFSArray<CFSWString> do_all(CFSWString utt, bool print_label, bool print_utt) {
CFSArray<CFSWString> res;
CFSArray<CPTWord> PTW;
utterance_struct u;
u.s = utt;
u.syl_c = 0;
u.phone_c = 0;
u.phra_c = do_phrases(u);
INTPTR word_count = 0;
if (print_utt) fprintf(stderr, "%s\n", ccstr(utt));
for (INTPTR i = 0; i < u.phr_vector.GetSize(); i++) {
u.phr_vector[i].utt_p = i;
phrase2words(u.phr_vector[i], PTW);
word_count += u.phr_vector[i].word_c;
}
CFSArray<CMorphInfos> MRs = Disambiguator.Disambiguate(Linguistic.AnalyzeSentense(PTW));
CFSArray<CMorphInfo> words;
for (INTPTR i = 0; i < MRs.GetSize(); i++)
//for (INTPTR i1 = 0; i1 < MRs[i].m_MorphInfo.GetSize(); i1++)
words.AddItem(MRs[i].m_MorphInfo[0]); //Ühestamistulemuse ühestamise koht
u.word_c = words.GetSize();
word_struct w;
INTPTR utt_phone_c = 1;
INTPTR syl_utt_p = 1;
INTPTR phone_utt_p = 1;
for (INTPTR i = 0; i < u.phr_vector.GetSize(); i++) {
u.phr_vector[i].utt_p = i + 1;
INTPTR syl_phr_p = 1;
INTPTR phone_phr_p = 1;
INTPTR phrase_pho_c = 1;
for (INTPTR i1 = 0; i1 < u.phr_vector[i].word_c; i1++) {
w.utt_p = utt_phone_c++;
w.phr_p = i1 + 1;
w.mi = words[0];
w.mi.m_szRoot += make_char_string(w.mi.m_szEnding) + w.mi.m_szClitic;
w.mi.m_szRoot = w.mi.m_szRoot.ToLower();
// sidesõnad + ei välteta
if ((CFSWString(w.mi.m_cPOS) == L"J") || (w.mi.m_szRoot == L"<ei")) w.mi.m_szRoot.Replace(L"<", L"", 1);
do_syls(w);
u.phr_vector[i].word_vector.AddItem(w);
INTPTR phone_word_p = 1;
INTPTR word_pho_c = 1;
for (INTPTR i2 = 0; i2 < u.phr_vector[i].word_vector[i1].syl_vector.GetSize(); i2++) {
u.syl_c++;
u.phr_vector[i].syl_c++;
u.phr_vector[i].word_vector[i1].syl_c++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].phr_p = syl_phr_p++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].utt_p = syl_utt_p++;
INTPTR syl_phone_c = 1;
do_phones(u.phr_vector[i].word_vector[i1].syl_vector[i2]);
for (INTPTR i3 = 0; i3 < u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector.GetSize(); i3++) {
u.phone_c++;
u.phr_vector[i].phone_c = phrase_pho_c++;
u.phr_vector[i].word_vector[i1].phone_c = word_pho_c++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_c = syl_phone_c++;
phone_struct p = u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector[i3];
p.utt_p = phone_utt_p++;
p.phr_p = phone_phr_p++;
p.word_p = phone_word_p++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector[i3] = p;
}
}
words.RemoveItem(0, 1);
}
}
if (print_label) print_u(u);
res = do_label(u);
return res;
}
