Anchor-annotator 0.9.0__tar.gz → 0.9.2__tar.gz

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2/Anchor_annotator.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: Anchor_annotator
-Version: 0.9.0
+Version: 0.9.2
 Summary: Anchor annotator is a program for inspecting corpora for the Montreal Forced Aligner and correcting transcriptions and pronunciations.
 Home-page: https://github.com/MontrealCorpusTools/Anchor-annotator
 Author: Montreal Corpus Tools

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/Anchor_annotator.egg-info/SOURCES.txt

@@ -271,6 +271,7 @@ docs/source/_static/img/utterance_unaligned.png
 docs/source/_static/img/utterance_waveform.png
 docs/source/_static/img/utterance_zoom.png
 docs/source/changelog/index.rst
+docs/source/first_steps/examples.rst
 docs/source/first_steps/index.rst
 docs/source/reference/index.rst
 docs/source/user_guide/configuration.rst

{anchor_annotator-0.9.0/Anchor_annotator.egg-info → anchor_annotator-0.9.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: Anchor_annotator
-Version: 0.9.0
+Version: 0.9.2
 Summary: Anchor annotator is a program for inspecting corpora for the Montreal Forced Aligner and correcting transcriptions and pronunciations.
 Home-page: https://github.com/MontrealCorpusTools/Anchor-annotator
 Author: Montreal Corpus Tools

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/anchor/_version.py

@@ -17,5 +17,5 @@ __version__: str
 __version_tuple__: VERSION_TUPLE
 version_tuple: VERSION_TUPLE
 
-__version__ = version = '0.9.0'
-__version_tuple__ = version_tuple = (0, 9, 0)
+__version__ = version = '0.9.2'
+__version_tuple__ = version_tuple = (0, 9, 2)

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/anchor/command_line.py

@@ -6,6 +6,9 @@ from montreal_forced_aligner.helper import configure_logger
 
 from anchor.main import Application, MainWindow
 
+if sys.platform == 'darwin':
+    os.environ["QT_MEDIA_BACKEND"] = "darwin"
+
 
 def main(debug=False):
     configure_logger("anchor")

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/anchor/main.py

@@ -764,9 +764,6 @@ class MainWindow(QtWidgets.QMainWindow):
         for worker in self.workers:
             worker.stopped.set()
         self.file_utterances_model.clean_up_for_close()
-        self.settings.setValue(
-            AnchorSettings.UTTERANCES_VISIBLE, self.ui.utteranceDockWidget.isVisible()
-        )
         self.settings.setValue(
             AnchorSettings.DICTIONARY_VISIBLE, self.ui.dictionaryDockWidget.isVisible()
         )
@@ -934,7 +931,6 @@ class MainWindow(QtWidgets.QMainWindow):
         self.corpus_undo_stack.cleanChanged.connect(self.corpus_changed)
         self.ui.lockEditAct.toggled.connect(self.undo_act.setDisabled)
         self.ui.lockEditAct.toggled.connect(self.redo_act.setDisabled)
-        self.ui.menuWindow.addAction(self.ui.utteranceDockWidget.toggleViewAction())
         self.ui.menuWindow.addAction(self.ui.dictionaryDockWidget.toggleViewAction())
         self.ui.menuWindow.addAction(self.ui.oovDockWidget.toggleViewAction())
         self.ui.menuWindow.addAction(self.ui.speakerDockWidget.toggleViewAction())
@@ -1217,6 +1213,7 @@ class MainWindow(QtWidgets.QMainWindow):
 
             for m in (
                 session.query(anchor.db.Dictionary)
+                .filter(anchor.db.Dictionary.path != '.')
                 .filter_by(available_locally=True)
                 .order_by(anchor.db.Dictionary.last_used.desc())
             ):
@@ -1839,7 +1836,7 @@ class MainWindow(QtWidgets.QMainWindow):
             self.ui.titleScreen.setVisible(False)
 
             self.ui.utteranceDockWidget.setVisible(
-                self.settings.value(AnchorSettings.UTTERANCES_VISIBLE)
+                True
             )
             self.ui.dictionaryDockWidget.setVisible(
                 self.settings.value(AnchorSettings.DICTIONARY_VISIBLE)
@@ -1972,8 +1969,6 @@ class MainWindow(QtWidgets.QMainWindow):
                 if dock_tab_bar.tabText(i) == "Utterances":
                     dock_tab_bar.setCurrentIndex(i)
                     break
-            else:
-                self.ui.utteranceDockWidget.toggleViewAction().trigger()
             self.ui.utteranceListWidget.search_box.setFocus()
             if search_term is not None:
                 self.ui.utteranceListWidget.search_box.setQuery(search_term)
@@ -1989,8 +1984,6 @@ class MainWindow(QtWidgets.QMainWindow):
                 if dock_tab_bar.tabText(i) == "Utterances":
                     dock_tab_bar.setCurrentIndex(i)
                     break
-            else:
-                self.ui.utteranceDockWidget.toggleViewAction().trigger()
 
     def open_search_speaker(self, search_term=None, show=False):
         if search_term is not None:

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/anchor/models.py

@@ -1375,11 +1375,18 @@ class DictionaryTableModel(TableModel):
         self.g2p_generator: typing.Optional[PyniniValidator] = None
         self.word_sets = {}
         self.speaker_mapping = {}
-        self.phones = []
+
         self.graphemes = []
         self.reference_phone_set = set()
         self.custom_mapping = {}
 
+    @property
+    def phones(self):
+        phones = [x.phone for x in self.corpus_model.phones.values() if x.phone_type == PhoneType.non_silence]
+        if self.corpus_model.corpus.position_dependent_phones:
+            phones = sorted(set(x.rsplit("_", maxsplit=1)[0] for x in phones))
+        return phones
+
     def set_custom_mapping(self, path):
         with mfa_open(path, "r") as f:
             self.custom_mapping = {k: v for k, v in yaml.safe_load(f).items() if k in self.phones}
@@ -1413,15 +1420,6 @@ class DictionaryTableModel(TableModel):
 
     def setup(self) -> None:
         self.refresh_dictionaries()
-        phones = [
-            x
-            for x, in self.corpus_model.session.query(Phone.phone).filter(
-                Phone.phone_type == PhoneType.non_silence
-            )
-        ]
-        if self.corpus_model.corpus.position_dependent_phones:
-            phones = sorted(set(x.rsplit("_", maxsplit=1)[0] for x in phones))
-        self.phones = phones
         specials = self.corpus_model.corpus.specials_set
         specials.update(
             [

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/anchor/plot.py

@@ -10,7 +10,8 @@ from typing import Optional
 import numpy as np
 import pyqtgraph as pg
 import sqlalchemy
-from Bio import pairwise2
+from kalpy.gmm.data import CtmInterval
+from _kalpy.util import align_intervals
 from montreal_forced_aligner.db import (
     PhoneInterval,
     Pronunciation,
@@ -600,6 +601,7 @@ class UtteranceView(QtWidgets.QWidget):
             tier.draggingLine.connect(self.audio_plot.update_drag_line)
             tier.lineDragFinished.connect(self.audio_plot.hide_drag_line)
             tier.receivedWheelEvent.connect(self.audio_plot.wheelEvent)
+            tier.receivedGestureEvent.connect(self.audio_plot.gestureEvent)
             tier.set_extra_tiers(self.extra_tiers)
             tier.setZValue(30)
             available_speakers[speaker_name] = speaker_id
@@ -1136,9 +1138,8 @@ class TranscriberErrorHighlighter(QtGui.QSyntaxHighlighter):
         if self.alignment:
             current_align_ind = 0
             for word_object in re.finditer(self.WORDS, text.lower()):
-                sb = self.alignment.seqB[current_align_ind]
-                sa = self.alignment.seqA[current_align_ind]
-                if sb == "-":
+                sa, sb = self.alignment[current_align_ind]
+                if sb.label == "-":
                     start = word_object.start() - 1
                     if start < 0:
                         start = 0
@@ -1148,19 +1149,18 @@ class TranscriberErrorHighlighter(QtGui.QSyntaxHighlighter):
                         count,
                         self.highlight_format,
                     )
-                    while sb != word_object.group():
+                    while sb.label != word_object.group():
                         current_align_ind += 1
-                        sb = self.alignment.seqB[current_align_ind]
-                        sa = self.alignment.seqA[current_align_ind]
-                if sb == word_object.group():
-                    if sb != sa:
+                        sa, sb = self.alignment[current_align_ind]
+                if sb.label == word_object.group():
+                    if sb.label != sa.label:
                         self.setFormat(
                             word_object.start(),
                             word_object.end() - word_object.start(),
                             self.highlight_format,
                         )
                     current_align_ind += 1
-            if current_align_ind < len(self.alignment.seqB):
+            if current_align_ind < len(self.alignment):
                 self.setFormat(
                     len(text) - 1,
                     1,
@@ -2652,16 +2652,9 @@ class UtteranceRegion(MfaRegion):
 
     def update_transcription_highlight(self):
         if self.item.normalized_text and self.item.transcription_text:
-            alignment = pairwise2.align.globalms(
-                self.item.normalized_text.lower().split(),
-                self.item.transcription_text.lower().split(),
-                0,
-                -2,
-                -1,
-                -1,
-                gap_char=["-"],
-                one_alignment_only=True,
-            )[0]
+            ref_intervals = [CtmInterval(0.0, 0.0, w) for w in self.item.normalized_text.lower().split()]
+            test_intervals = [CtmInterval(0.0, 0.0, w) for w in self.item.transcription_text.lower().split()]
+            alignment = align_intervals(ref_intervals, test_intervals, "<eps>", {})
             self.transcription_text.highlighter.set_alignment(alignment)
 
     def set_search_term(self, term):
@@ -3264,6 +3257,7 @@ class AudioPlots(pg.GraphicsObject):
         self.wave_form.setParentItem(self)
         self.spectrogram.setParentItem(self)
         self.pitch_track.setParentItem(self)
+        self.grabGesture(QtCore.Qt.PinchGesture)
         color = self.plot_theme.selected_range_color
         color.setAlphaF(0.25)
         self.selection_brush = pg.mkBrush(color)
@@ -3318,6 +3312,20 @@ class AudioPlots(pg.GraphicsObject):
     def hide_drag_line(self):
         self.update_line.hide()
 
+    def sceneEvent(self, ev):
+        if ev.type() == QtCore.QEvent.Gesture:
+            return self.gestureEvent(ev)
+        return super().sceneEvent(ev)
+
+    def gestureEvent(self, ev):
+        ev.accept()
+        pinch = ev.gesture(QtCore.Qt.PinchGesture)
+        if pinch is not None:
+            delta = pinch.scaleFactor()
+            sc = delta
+            center = self.getViewBox().mapToView(pinch.centerPoint())
+            self.selection_model.zoom(sc, center.x())
+
     def wheelEvent(self, ev: QtWidgets.QGraphicsSceneWheelEvent):
         ev.accept()
         delta = ev.delta()
@@ -3446,6 +3454,7 @@ class AudioPlots(pg.GraphicsObject):
 
 class SpeakerTier(pg.GraphicsObject):
     receivedWheelEvent = QtCore.Signal(object)
+    receivedGestureEvent = QtCore.Signal(object)
     draggingLine = QtCore.Signal(object)
     lineDragFinished = QtCore.Signal(object)
 
@@ -3499,10 +3508,22 @@ class SpeakerTier(pg.GraphicsObject):
         self.selection_model.model().utterancesReady.connect(self.refresh)
         self.available_speakers = {}
         self.painter_path_cache = {}
+        self.grabGesture(QtCore.Qt.PinchGesture)
 
     def wheelEvent(self, ev):
         self.receivedWheelEvent.emit(ev)
 
+    def sceneEvent(self, ev):
+        if ev.type() == QtCore.QEvent.Gesture:
+            return self.gestureEvent(ev)
+        return super().sceneEvent(ev)
+
+    def gestureEvent(self, ev):
+        ev.accept()
+        pinch = ev.gesture(QtCore.Qt.PinchGesture)
+        if pinch is not None:
+            self.receivedGestureEvent.emit(ev)
+
     def create_utterance(self, begin, end):
         self.file_model.create_utterance(self.speaker_id, begin, end)
 

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/anchor/undo.py

@@ -448,12 +448,15 @@ class DeleteReferenceIntervalsCommand(FileCommand):
         super().__init__(file_model)
         self.utterance = utterance
         self.reference_intervals = None
+        self.reference_word_intervals = None
         self.reference_workflow = None
 
     def _redo(self, session) -> None:
         if self.reference_intervals is None:
             self.reference_intervals = self.utterance.reference_phone_intervals
+            self.reference_word_intervals = self.utterance.reference_word_intervals
         self.utterance.reference_phone_intervals = []
+        self.utterance.reference_word_intervals = []
         self.utterance.manual_alignments = False
         session.merge(self.utterance)
 
@@ -462,10 +465,17 @@ class DeleteReferenceIntervalsCommand(FileCommand):
         for pi in self.reference_intervals:
             make_transient(pi)
             reference_phone_intervals.append(pi)
+        reference_word_intervals = []
+        for wi in self.reference_word_intervals:
+            make_transient(wi)
+            reference_word_intervals.append(wi)
         self.utterance.manual_alignments = True
         self.utterance.reference_phone_intervals = sorted(
             reference_phone_intervals, key=lambda x: x.begin
         )
+        self.utterance.reference_word_intervals = sorted(
+            reference_word_intervals, key=lambda x: x.begin
+        )
         session.merge(self.utterance)
 
     def update_data(self):

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/anchor/widgets.py

@@ -1732,7 +1732,7 @@ class IpaKeyboard(QtWidgets.QMenu):
     def __init__(self, phones, parent=None):
         super().__init__(parent)
         self.settings = AnchorSettings()
-        layout = QtWidgets.QVBoxLayout()
+        layout = QtWidgets.QHBoxLayout()
         self.scroll_area = QtWidgets.QScrollArea(self)
         self.scroll_area.setFocusPolicy(QtCore.Qt.FocusPolicy.NoFocus)
         widget = QtWidgets.QWidget(self)
@@ -1753,12 +1753,13 @@ class IpaKeyboard(QtWidgets.QMenu):
                 col_index = 0
                 row_index += 1
         layout.addWidget(self.scroll_area)
+        scroll_layout.setContentsMargins(0,0,0,0)
         widget.setLayout(scroll_layout)
         self.scroll_area.setWidget(widget)
         self.setLayout(layout)
         self.setFocusPolicy(QtCore.Qt.FocusPolicy.NoFocus)
         self.scroll_area.setMinimumWidth(
-            widget.sizeHint().width() + self.scroll_area.verticalScrollBar().sizeHint().width()
+            widget.sizeHint().width() + self.scroll_area.verticalScrollBar().sizeHint().width() + 1
         )
         self.scroll_area.setHorizontalScrollBarPolicy(QtCore.Qt.ScrollBarPolicy.ScrollBarAlwaysOff)
         self.setStyleSheet(self.settings.keyboard_style_sheet)
@@ -1773,7 +1774,10 @@ class IpaKeyboard(QtWidgets.QMenu):
         self.inputPhone.emit(b.text(), True)
 
     def showEvent(self, event: QtGui.QShowEvent) -> None:
-        p = self.pos()
+        super().showEvent(event)
+        pos = self.parent().geometry().bottomLeft()
+        p = self.parent().mapToGlobal(pos)
+
         geo = self.geometry()
         new_pos = int(p.x() - (geo.width() / 2))
         self.move(new_pos, p.y())
@@ -1814,6 +1818,21 @@ class PronunciationField(QtWidgets.QTextEdit):
         self.setHorizontalScrollBarPolicy(QtCore.Qt.ScrollBarPolicy.ScrollBarAlwaysOff)
 
 
+class KeyboardWidget(QtWidgets.QPushButton):
+    def __init__(self, phones, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.setFocusPolicy(QtCore.Qt.FocusPolicy.NoFocus)
+        keyboard_icon = QtGui.QIcon.fromTheme("input-keyboard")
+        self.setIcon(keyboard_icon)
+        self.keyboard = IpaKeyboard(phones, self)
+        self.setMenu(self.keyboard)
+
+        self.clicked.connect(self.showMenu)
+        
+    def showMenu(self):
+        self.menu().show()
+
+
 class PronunciationInput(QtWidgets.QToolBar):
     validationError = QtCore.Signal(object)
     returnPressed = QtCore.Signal()
@@ -1842,15 +1861,10 @@ class PronunciationInput(QtWidgets.QToolBar):
 
         self.cancel_action = QtGui.QAction(icon=cancel_icon, parent=self)
         self.cancel_action.triggered.connect(self.cancel)
-        keyboard_icon = QtGui.QIcon.fromTheme("input-keyboard")
 
-        self.keyboard_widget = QtWidgets.QPushButton(self)
-        self.keyboard_widget.setFocusPolicy(QtCore.Qt.FocusPolicy.NoFocus)
-        self.keyboard_widget.setIcon(keyboard_icon)
-        self.keyboard = IpaKeyboard(phones)
-        self.keyboard.installEventFilter(self)
-        self.keyboard.inputPhone.connect(self.add_phone)
-        self.keyboard_widget.setMenu(self.keyboard)
+        self.keyboard_widget = KeyboardWidget(phones, self)
+        self.keyboard_widget.keyboard.installEventFilter(self)
+        self.keyboard_widget.keyboard.inputPhone.connect(self.add_phone)
 
         self.addWidget(self.input)
         self.addWidget(self.keyboard_widget)
@@ -2033,14 +2047,14 @@ class EditableDelegate(QtWidgets.QStyledItemDelegate):
 
     def setEditorData(
         self,
-        editor: PronunciationInput,
+        editor: WordInput,
         index: typing.Union[QtCore.QModelIndex, QtCore.QPersistentModelIndex],
     ) -> None:
         editor.setText(index.model().data(index, QtCore.Qt.ItemDataRole.EditRole))
 
     def setModelData(
         self,
-        editor: PronunciationInput,
+        editor: WordInput,
         model: DictionaryTableModel,
         index: typing.Union[QtCore.QModelIndex, QtCore.QPersistentModelIndex],
     ) -> None:
@@ -2051,7 +2065,7 @@ class EditableDelegate(QtWidgets.QStyledItemDelegate):
 
     def updateEditorGeometry(
         self,
-        editor: PronunciationInput,
+        editor: WordInput,
         option: QtWidgets.QStyleOptionViewItem,
         index: typing.Union[QtCore.QModelIndex, QtCore.QPersistentModelIndex],
     ) -> None:

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/anchor/workers.py

@@ -55,6 +55,7 @@ from montreal_forced_aligner.db import (
     Dictionary2Job,
     File,
     Phone,
+    Job,
     PhoneInterval,
     Pronunciation,
     ReferencePhoneInterval,
@@ -3935,6 +3936,11 @@ class ImportDictionaryWorker(FunctionWorker):  # pragma: no cover
                 session.query(Word).delete()
                 session.query(Phone).delete()
                 session.execute(sqlalchemy.update(Speaker).values(dictionary_id=None))
+                session.flush()
+                d = session.query(Dictionary).filter(Dictionary.name=='.').first()
+                if d is not None:
+                    d.delete()
+                session.flush()
                 session.execute(
                     sqlalchemy.update(CorpusWorkflow).values(
                         done=False, alignments_collected=False, score=None
@@ -3945,8 +3951,12 @@ class ImportDictionaryWorker(FunctionWorker):  # pragma: no cover
                 session.commit()
             corpus.dictionary_setup()
             with corpus.session() as session:
+                d = session.query(Dictionary).filter(Dictionary.name==self.dictionary_path.stem).first()
+                jobs = session.query(Job).all()
+                for j in jobs:
+                    j.dictionaries = [d]
                 session.execute(
-                    sqlalchemy.update(Speaker).values(dictionary_id=corpus._default_dictionary_id)
+                    sqlalchemy.update(Speaker).values(dictionary_id=d.id)
                 )
                 session.commit()
             corpus.text_normalized = False

anchor_annotator-0.9.2/docs/source/first_steps/examples.rst

@@ -0,0 +1,39 @@
+
+.. _`LibriSpeech data set`: https://drive.google.com/open?id=1MNlwIv5VyMemrXcZCcC6hENSZpojkdpm
+
+.. _examples:
+
+********
+Examples
+********
+
+.. _alignment_example:
+
+Example 1: Aligning LibriSpeech (English)
+=========================================
+
+.. note::
+
+   This example mirrors the example in
+   `the MFA alignment example <https://montreal-forced-aligner.readthedocs.io/en/latest/first_steps/example.html#example-1-aligning-librispeech-english>`_.
+
+Set up
+------
+
+1. Ensure you have installed Anchor via :ref:`installation`.
+4. Download the prepared LibriSpeech dataset (`LibriSpeech data set`_) and extract it somewhere on your computer
+
+
+Loading the corpus
+------------------
+
+In a terminal with the conda environment created in :ref:`installation` activated enter the following:
+
+.. code-block:: bash
+
+   mfa anchor
+
+.. warning::
+
+   Under construction
+

{anchor_annotator-0.9.0 → anchor_annotator-0.9.2}/requirements.txt

@@ -12,5 +12,4 @@ pyyaml
 scipy
 scikit-learn
 pynini
-biopython
 kalpy