static void Table_setEventTypeString (Table me) {
try {
for (long i = 1; i <= my rows.size; i ++) {
int type = Table_getNumericValue_Assert (me, i, 2);
const char32 *label = U"0";
if (type == espeakEVENT_WORD) {
label = U"word";
} else if (type == espeakEVENT_SENTENCE) {
label = U"sent";
} else if (type == espeakEVENT_MARK) {
label = U"mark";
} else if (type == espeakEVENT_PLAY) {
label = U"play";
} else if (type == espeakEVENT_END) {
label = U"s-end";
} else if (type == espeakEVENT_MSG_TERMINATED) {
label = U"msg_term";
} else if (type == espeakEVENT_PHONEME) {
label = U"phoneme";
}
Table_setStringValue (me, i, 3, label);
}
} catch (MelderError) {
Melder_throw (U"Event types not set.");
}
}
LinearRegression Table_to_LinearRegression (Table me) {
try {
long numberOfIndependentVariables = my numberOfColumns - 1, numberOfParameters = my numberOfColumns;
long numberOfCells = my rows -> size, icell, ivar;
if (numberOfParameters < 1) /* Includes intercept. */
Melder_throw (U"Not enough columns (has to be more than 1).");
if (numberOfCells < numberOfParameters) {
Melder_warning (U"Solution is not unique (more parameters than cases).");
}
autoNUMmatrix <double> u (1, numberOfCells, 1, numberOfParameters);
autoNUMvector <double> b (1, numberOfCells);
autoNUMvector <double> x (1, numberOfParameters);
autoLinearRegression thee = LinearRegression_create ();
for (ivar = 1; ivar <= numberOfIndependentVariables; ivar ++) {
double minimum = Table_getMinimum (me, ivar);
double maximum = Table_getMaximum (me, ivar);
Regression_addParameter (thee.peek(), my columnHeaders [ivar]. label, minimum, maximum, 0.0);
}
for (icell = 1; icell <= numberOfCells; icell ++) {
for (ivar = 1; ivar < numberOfParameters; ivar ++) {
u [icell] [ivar] = Table_getNumericValue_Assert (me, icell, ivar);
}
u [icell] [numberOfParameters] = 1.0; /* For the intercept. */
b [icell] = Table_getNumericValue_Assert (me, icell, my numberOfColumns); // the dependent variable
}
NUMsolveEquation (u.peek(), numberOfCells, numberOfParameters, b.peek(), NUMeps * numberOfCells, x.peek());
thy intercept = x [numberOfParameters];
for (ivar = 1; ivar <= numberOfIndependentVariables; ivar ++) {
RegressionParameter parm = static_cast<RegressionParameter> (thy parameters -> item [ivar]);
parm -> value = x [ivar];
}
return thee.transfer();
} catch (MelderError) {
Melder_throw (me, U": linear regression not performed.");
}
}
static LogisticRegression _Table_to_LogisticRegression (Table me, long *factors, long numberOfFactors, long dependent1, long dependent2) {
long numberOfParameters = numberOfFactors + 1;
long numberOfCells = my rows -> size, numberOfY0 = 0, numberOfY1 = 0, numberOfData = 0;
double logLikelihood = 1e300, previousLogLikelihood = 2e300;
if (numberOfParameters < 1) // includes intercept
Melder_throw ("Not enough columns (has to be more than 1).");
/*
* Divide up the contents of the table into a number of independent variables (x) and two dependent variables (y0 and y1).
*/
autoNUMmatrix <double> x (1, numberOfCells, 0, numberOfFactors); // column 0 is the intercept
autoNUMvector <double> y0 (1, numberOfCells);
autoNUMvector <double> y1 (1, numberOfCells);
autoNUMvector <double> meanX (1, numberOfFactors);
autoNUMvector <double> stdevX (1, numberOfFactors);
autoNUMmatrix <double> smallMatrix (0, numberOfFactors, 0, numberOfParameters);
autoLogisticRegression thee = LogisticRegression_create (my columnHeaders [dependent1]. label, my columnHeaders [dependent2]. label);
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
double minimum = Table_getMinimum (me, factors [ivar]);
double maximum = Table_getMaximum (me, factors [ivar]);
Regression_addParameter (thee.peek(), my columnHeaders [factors [ivar]]. label, minimum, maximum, 0.0);
}
for (long icell = 1; icell <= numberOfCells; icell ++) {
y0 [icell] = Table_getNumericValue_Assert (me, icell, dependent1);
y1 [icell] = Table_getNumericValue_Assert (me, icell, dependent2);
numberOfY0 += y0 [icell];
numberOfY1 += y1 [icell];
numberOfData += y0 [icell] + y1 [icell];
x [icell] [0] = 1.0; /* Intercept. */
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
x [icell] [ivar] = Table_getNumericValue_Assert (me, icell, factors [ivar]);
meanX [ivar] += x [icell] [ivar] * (y0 [icell] + y1 [icell]);
}
}
if (numberOfY0 == 0 && numberOfY1 == 0)
Melder_throw ("No data in either class. Cannot determine result.");
if (numberOfY0 == 0)
Melder_throw ("No data in class ", my columnHeaders [dependent1]. label, ". Cannot determine result.");
if (numberOfY1 == 0)
Melder_throw ("No data in class ", my columnHeaders [dependent2]. label, ". Cannot determine result.");
/*
* Normalize the data.
*/
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
meanX [ivar] /= numberOfData;
for (long icell = 1; icell <= numberOfCells; icell ++) {
x [icell] [ivar] -= meanX [ivar];
}
}
for (long icell = 1; icell <= numberOfCells; icell ++) {
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
stdevX [ivar] += x [icell] [ivar] * x [icell] [ivar] * (y0 [icell] + y1 [icell]);
}
}
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
stdevX [ivar] = sqrt (stdevX [ivar] / numberOfData);
for (long icell = 1; icell <= numberOfCells; icell ++) {
x [icell] [ivar] /= stdevX [ivar];
}
}
/*
* Initial state of iteration: the null model.
*/
thy intercept = log ((double) numberOfY1 / (double) numberOfY0); // initial state of intercept: best guess for average log odds
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
RegressionParameter parm = static_cast<RegressionParameter> (thy parameters -> item [ivar]);
parm -> value = 0.0; // initial state of dependence: none
}
long iteration = 1;
for (; iteration <= 100; iteration ++) {
previousLogLikelihood = logLikelihood;
for (long ivar = 0; ivar <= numberOfFactors; ivar ++) {
for (long jvar = ivar; jvar <= numberOfParameters; jvar ++) {
smallMatrix [ivar] [jvar] = 0.0;
}
}
/*
* Compute the current log likelihood.
*/
logLikelihood = 0.0;
for (long icell = 1; icell <= numberOfCells; icell ++) {
double fittedLogit = thy intercept, fittedP, fittedQ, fittedLogP, fittedLogQ, fittedPQ, fittedVariance;
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
RegressionParameter parm = static_cast<RegressionParameter> (thy parameters -> item [ivar]);
fittedLogit += parm -> value * x [icell] [ivar];
}
/*
* Basically we have fittedP = 1.0 / (1.0 + exp (- fittedLogit)),
* but that works neither for fittedP values near 0 nor for values near 1.
*/
if (fittedLogit > 15.0) {
/*
* For large fittedLogit, fittedLogP = ln (1/(1+exp(-fittedLogit))) = -ln (1+exp(-fittedLogit)) =~ - exp(-fittedLogit)
*/
fittedLogP = - exp (- fittedLogit);
fittedLogQ = - fittedLogit;
fittedPQ = exp (- fittedLogit);
fittedP = exp (fittedLogP);
fittedQ = 1.0 - fittedP;
} else if (fittedLogit < -15.0) {
fittedLogP = fittedLogit;
fittedLogQ = - exp (fittedLogit);
fittedPQ = exp (fittedLogit);
fittedP = exp (fittedLogP);
fittedQ = 1 - fittedP;
} else {
fittedP = 1.0 / (1.0 + exp (- fittedLogit));
fittedLogP = log (fittedP);
fittedQ = 1.0 - fittedP;
fittedLogQ = log (fittedQ);
fittedPQ = fittedP * fittedQ;
}
logLikelihood += -2 * (y1 [icell] * fittedLogP + y0 [icell] * fittedLogQ);
/*
* Matrix shifting stuff.
* Suppose a + b Sk + c Tk = ln (pk / qk),
* where {a, b, c} are the coefficients to be optimized,
* Sk and Tk are properties of stimulus k,
* and pk and qk are the fitted probabilities for y1 and y0, respectively, given stimulus k.
* Then ln pk = - ln (1 + qk / pk) = - ln (1 + exp (- (a + b Sk + c Tk)))
* d ln pk / da = 1 / (1 + exp (a + b Sk + c Tk)) = qk
* d ln pk / db = qk Sk
* d ln pk / dc = qk Tk
* d ln qk / da = - pk
* Now LL = Sum(k) (y1k ln pk + y0k ln qk)
* so that dLL/da = Sum(k) (y1k d ln pk / da + y0k ln qk / da) = Sum(k) (y1k qk - y0k pk)
*/
fittedVariance = fittedPQ * (y0 [icell] + y1 [icell]);
for (long ivar = 0; ivar <= numberOfFactors; ivar ++) {
/*
* The last column gets the gradient of LL: dLL/da, dLL/db, dLL/dc.
*/
smallMatrix [ivar] [numberOfParameters] += x [icell] [ivar] * (y1 [icell] * fittedQ - y0 [icell] * fittedP);
for (long jvar = ivar; jvar <= numberOfFactors; jvar ++) {
smallMatrix [ivar] [jvar] += x [icell] [ivar] * x [icell] [jvar] * fittedVariance;
}
}
}
if (fabs (logLikelihood - previousLogLikelihood) < 1e-11) {
break;
}
/*
* Make matrix symmetric.
*/
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
for (long jvar = 0; jvar < ivar; jvar ++) {
smallMatrix [ivar] [jvar] = smallMatrix [jvar] [ivar];
}
}
/*
* Invert matrix in the simplest way, and shift and wipe the last column with it.
*/
for (long ivar = 0; ivar <= numberOfFactors; ivar ++) {
double pivot = smallMatrix [ivar] [ivar]; /* Save diagonal. */
smallMatrix [ivar] [ivar] = 1.0;
for (long jvar = 0; jvar <= numberOfParameters; jvar ++) {
smallMatrix [ivar] [jvar] /= pivot;
}
for (long jvar = 0; jvar <= numberOfFactors; jvar ++) {
if (jvar != ivar) {
double temp = smallMatrix [jvar] [ivar];
smallMatrix [jvar] [ivar] = 0.0;
for (long kvar = 0; kvar <= numberOfParameters; kvar ++) {
smallMatrix [jvar] [kvar] -= temp * smallMatrix [ivar] [kvar];
}
}
}
}
/*
* Update the parameters from the last column of smallMatrix.
*/
thy intercept += smallMatrix [0] [numberOfParameters];
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
RegressionParameter parm = static_cast<RegressionParameter> (thy parameters -> item [ivar]);
parm -> value += smallMatrix [ivar] [numberOfParameters];
}
}
if (iteration > 100) {
Melder_warning (L"Logistic regression has not converged in 100 iterations. The results are unreliable.");
}
for (long ivar = 1; ivar <= numberOfFactors; ivar ++) {
RegressionParameter parm = static_cast<RegressionParameter> (thy parameters -> item [ivar]);
parm -> value /= stdevX [ivar];
thy intercept -= parm -> value * meanX [ivar];
}
return thee.transfer();
}
autoSound SpeechSynthesizer_to_Sound (SpeechSynthesizer me, const char32 *text, autoTextGrid *tg, autoTable *events) {
try {
int fsamp = espeak_Initialize (AUDIO_OUTPUT_SYNCHRONOUS, 0, nullptr, // 5000ms
espeakINITIALIZE_PHONEME_EVENTS|espeakINITIALIZE_PHONEME_IPA);
if (fsamp == -1) {
Melder_throw (U"Internal espeak error.");
}
int synth_flags = espeakCHARS_WCHAR;
if (my d_inputTextFormat == SpeechSynthesizer_INPUT_TAGGEDTEXT) {
synth_flags |= espeakSSML;
}
if (my d_inputTextFormat != SpeechSynthesizer_INPUT_TEXTONLY) {
synth_flags |= espeakPHONEMES;
}
option_phoneme_events = espeakINITIALIZE_PHONEME_EVENTS; // extern int option_phoneme_events;
if (my d_outputPhonemeCoding == SpeechSynthesizer_PHONEMECODINGS_IPA) {
option_phoneme_events |= espeakINITIALIZE_PHONEME_IPA;
}
espeak_SetParameter (espeakRATE, my d_wordsPerMinute, 0);
espeak_SetParameter (espeakPITCH, my d_pitchAdjustment, 0);
espeak_SetParameter (espeakRANGE, my d_pitchRange, 0);
const char32 *voiceLanguageCode = SpeechSynthesizer_getVoiceLanguageCodeFromName (me, my d_voiceLanguageName);
const char32 *voiceVariantCode = SpeechSynthesizer_getVoiceVariantCodeFromName (me, my d_voiceVariantName);
espeakdata_SetVoiceByName ((const char *) Melder_peek32to8 (voiceLanguageCode),
(const char *) Melder_peek32to8 (voiceVariantCode));
espeak_SetParameter (espeakWORDGAP, my d_wordgap * 100, 0); // espeak wordgap is in units of 10 ms
espeak_SetParameter (espeakCAPITALS, 0, 0);
espeak_SetParameter (espeakPUNCTUATION, espeakPUNCT_NONE, 0);
espeak_SetSynthCallback (synthCallback);
my d_events = Table_createWithColumnNames (0, U"time type type-t t-pos length a-pos sample id uniq");
#ifdef _WIN32
wchar_t *textW = Melder_peek32toW (text);
espeak_Synth (textW, wcslen (textW) + 1, 0, POS_CHARACTER, 0, synth_flags, nullptr, me);
#else
espeak_Synth (text, str32len (text) + 1, 0, POS_CHARACTER, 0, synth_flags, nullptr, me);
#endif
espeak_Terminate ();
autoSound thee = buffer_to_Sound (my d_wav, my d_numberOfSamples, my d_internalSamplingFrequency);
if (my d_samplingFrequency != my d_internalSamplingFrequency) {
thee = Sound_resample (thee.get(), my d_samplingFrequency, 50);
}
my d_numberOfSamples = 0; // re-use the wav-buffer
if (tg) {
double xmin = Table_getNumericValue_Assert (my d_events.get(), 1, 1);
if (xmin > thy xmin) {
xmin = thy xmin;
}
double xmax = Table_getNumericValue_Assert (my d_events.get(), my d_events -> rows.size, 1);
if (xmax < thy xmax) {
xmax = thy xmax;
}
autoTextGrid tg1 = Table_to_TextGrid (my d_events.get(), text, xmin, xmax);
*tg = TextGrid_extractPart (tg1.get(), thy xmin, thy xmax, 0);
}
if (events) {
Table_setEventTypeString (my d_events.get());
*events = my d_events.move();
}
my d_events.reset();
return thee;
} catch (MelderError) {
espeak_Terminate ();
Melder_throw (U"Text not played.");
}
}
static autoTextGrid Table_to_TextGrid (Table me, const char32 *text, double xmin, double xmax) {
//Table_createWithColumnNames (0, L"time type type-t t-pos length a-pos sample id uniq");
try {
long length, textLength = str32len (text);
long numberOfRows = my rows.size;
long timeColumnIndex = Table_getColumnIndexFromColumnLabel (me, U"time");
long typeColumnIndex = Table_getColumnIndexFromColumnLabel (me, U"type");
long tposColumnIndex = Table_getColumnIndexFromColumnLabel (me, U"t-pos");
long idColumnIndex = Table_getColumnIndexFromColumnLabel (me, U"id");
autoTextGrid thee = TextGrid_create (xmin, xmax, U"sentence clause word phoneme", U"");
TextGrid_setIntervalText (thee.get(), 1, 1, text);
long p1c = 1, p1w = 1;
double time_phon_p = xmin;
bool wordEnd = false;
autoMelderString mark;
IntervalTier clauses = (IntervalTier) thy tiers->at [2];
IntervalTier words = (IntervalTier) thy tiers->at [3];
IntervalTier phonemes = (IntervalTier) thy tiers->at [4];
for (long i = 1; i <= numberOfRows; i++) {
double time = Table_getNumericValue_Assert (me, i, timeColumnIndex);
int type = Table_getNumericValue_Assert (me, i, typeColumnIndex);
long pos = Table_getNumericValue_Assert (me, i, tposColumnIndex);
if (type == espeakEVENT_SENTENCE) {
// Only insert a new boundary, no text
// text will be inserted at end sentence event
if (time > xmin && time < xmax) {
IntervalTier_addBoundaryUnsorted (clauses, clauses -> intervals.size, time, U"", true);
}
p1c = pos;
} else if (type == espeakEVENT_END) {
// End of clause: insert new boundary, and fill left interval with text
length = pos - p1c + 1;
MelderString_ncopy (&mark, text + p1c - 1, length);
MelderString_trimWhiteSpaceAtEnd (& mark);
if (time > xmin && time < xmax) {
IntervalTier_addBoundaryUnsorted (clauses, clauses -> intervals.size, time, mark.string, true);
} else {
TextGrid_setIntervalText (thee.get(), 2, clauses -> intervals.size, mark.string);
}
p1c = pos;
// End of clause always signals "end of a word"
if (pos <= textLength) {
length = pos - p1w + 1;
MelderString_ncopy (&mark, text + p1w - 1, length);
MelderString_trimWhiteSpaceAtEnd (& mark);
if (time > xmin && time < xmax) {
IntervalTier_addBoundaryUnsorted (words, words -> intervals.size, time, mark.string, true);
} else {
TextGrid_setIntervalText (thee.get(), 3, words -> intervals.size, mark.string);
}
// now the next word event should not trigger setting the left interval text
wordEnd = false;
}
} else if (type == espeakEVENT_WORD) {
if (pos < p1w) {
continue;
}
if (time > xmin && time < xmax) {
length = pos - p1w;
if (pos == textLength) {
length++;
}
MelderString_ncopy (&mark, text + p1w - 1, length);
MelderString_trimWhiteSpaceAtEnd (& mark);
IntervalTier_addBoundaryUnsorted (words, words -> intervals.size, time, ( wordEnd ? mark.string : U"" ), true);
}
wordEnd = true;
p1w = pos;
} else if (type == espeakEVENT_PHONEME) {
const char32 *id = Table_getStringValue_Assert (me, i, idColumnIndex);
if (time > time_phon_p) {
// Insert new boudary and label interval with the id
// TODO: Translate the id to the correct notation
TextInterval ti = phonemes -> intervals.at [phonemes -> intervals.size];
if (time > ti -> xmin && time < ti -> xmax) {
IntervalTier_addBoundaryUnsorted (phonemes, phonemes -> intervals.size, time, id, false);
}
} else {
// Just in case the phoneme starts at xmin we only need to set interval text
TextGrid_setIntervalText (thee.get(), 4, phonemes -> intervals.size, id);
}
time_phon_p = time;
}
}
clauses -> intervals. sort ();
words -> intervals. sort ();
phonemes -> intervals. sort ();
IntervalTier_mergeSpecialIntervals (phonemes); // Merge neighbouring empty U"" and U"\001" intervals
IntervalTier_removeVeryShortIntervals (words);
IntervalTier_removeVeryShortIntervals (clauses);
/* Use empty intervals in phoneme tier for more precision in the word tier */
IntervalTier_insertEmptyIntervalsFromOtherTier (words, phonemes);
IntervalTier_mergeSpecialIntervals (words); // Merge neighbouring empty U"" and U"\001" intervals
return thee;
} catch (MelderError) {
Melder_throw (U"TextGrid not created from Table with events.");
}
}
static autoTextGrid Table_to_TextGrid (Table me, const char32 *text, double xmin, double xmax) {
//Table_createWithColumnNames (0, L"time type type-t t-pos length a-pos sample id uniq");
try {
long length, textLength = str32len (text);
long numberOfRows = my rows -> size;
long timeColumnIndex = Table_getColumnIndexFromColumnLabel (me, U"time");
long typeColumnIndex = Table_getColumnIndexFromColumnLabel (me, U"type");
long tposColumnIndex = Table_getColumnIndexFromColumnLabel (me, U"t-pos");
long idColumnIndex = Table_getColumnIndexFromColumnLabel (me, U"id");
autoTextGrid thee = TextGrid_create (xmin, xmax, U"sentence clause word phoneme", U"");
TextGrid_setIntervalText (thee.peek(), 1, 1, text);
long p1c = 1, p1w = 1;
double t1p = xmin;
bool wordEnd = false;
autoMelderString mark;
IntervalTier itc = (IntervalTier) thy tiers -> item[2];
IntervalTier itw = (IntervalTier) thy tiers -> item[3];
IntervalTier itp = (IntervalTier) thy tiers -> item[4];
for (long i = 1; i <= numberOfRows; i++) {
double time = Table_getNumericValue_Assert (me, i, timeColumnIndex);
int type = Table_getNumericValue_Assert (me, i, typeColumnIndex);
long pos = Table_getNumericValue_Assert (me, i, tposColumnIndex);
if (type == espeakEVENT_SENTENCE) {
// Only insert a new boundary, no text
// text will be inserted at end sentence event
if (time > xmin and time < xmax) {
IntervalTier_addBoundaryUnsorted (itc, itc -> intervals -> size, time, U"", true);
}
p1c = pos;
} else if (type == espeakEVENT_END) {
// End of clause: insert new boundary, and fill left interval with text
length = pos - p1c + 1;
MelderString_ncopy (&mark, text + p1c - 1, length);
MelderString_trimWhiteSpaceAtEnd (&mark);
if (time > xmin and time < xmax) {
IntervalTier_addBoundaryUnsorted (itc, itc -> intervals -> size, time, mark.string, true);
} else {
TextGrid_setIntervalText (thee.peek(), 2, itc -> intervals -> size, mark.string);
}
p1c = pos;
// End of clause always signals "end of a word"
if (pos <= textLength) {
length = pos - p1w + 1;
MelderString_ncopy (&mark, text + p1w - 1, length);
MelderString_trimWhiteSpaceAtEnd (&mark);
if (time > xmin and time < xmax) {
IntervalTier_addBoundaryUnsorted (itw, itw -> intervals -> size, time, mark.string, true);
} else {
TextGrid_setIntervalText (thee.peek(), 3, itw -> intervals -> size, mark.string);
}
// now the next word event should not trigger setting the left interval text
wordEnd = false;
}
} else if (type == espeakEVENT_WORD) {
if (pos < p1w) {
continue;
}
if (time > xmin and time < xmax) {
length = pos - p1w;
if (pos == textLength) length++;
MelderString_ncopy (&mark, text + p1w - 1, length);
MelderString_trimWhiteSpaceAtEnd (&mark);
IntervalTier_addBoundaryUnsorted (itw, itw -> intervals -> size, time, (wordEnd ? mark.string : U""), true);
}
wordEnd = true;
p1w = pos;
} else if (type == espeakEVENT_PHONEME) {
const char32 *id = Table_getStringValue_Assert (me, i, idColumnIndex);
if (time > t1p) {
// Insert new boudary and label interval with the id
// TODO: Translate the id to the correct notation
TextInterval ti = (TextInterval) itp -> intervals -> item[itp -> intervals -> size];
if (time > ti -> xmin and time < ti -> xmax) {
IntervalTier_addBoundaryUnsorted (itp, itp -> intervals -> size, time, id, false);
}
} else {
// Just in case the phoneme starts at xmin we only need to set interval text
TextGrid_setIntervalText (thee.peek(), 4, itp -> intervals -> size, id);
}
t1p = time;
}
}
Sorted_sort (itc -> intervals);
Sorted_sort (itw -> intervals);
Sorted_sort (itp -> intervals);
return thee;
} catch (MelderError) {
Melder_throw (U"TextGrid not created from Table with events.");
}
}
