fix some bugs and fit ocr with low quality films

assemble_dialog.ui

@@ -13,15 +13,40 @@
   <property name="windowTitle">
    <string>Dialog</string>
   </property>
-  <widget class="QDialogButtonBox" name="buttonBox">
-   <property name="geometry">
-    <rect>
-     <x>220</x>
-     <y>250</y>
-     <width>341</width>
-     <height>32</height>
-    </rect>
+  <layout class="QGridLayout" name="gridLayout">
+   <item row="0" column="2">
+    <widget class="QPushButton" name="pushButton_2">
+     <property name="text">
+      <string>表格路径</string>
+     </property>
+    </widget>
+   </item>
+   <item row="1" column="1">
+    <widget class="QLineEdit" name="lineEdit_3">
+     <property name="enabled">
+      <bool>false</bool>
      </property>
+    </widget>
+   </item>
+   <item row="2" column="1">
+    <widget class="QLineEdit" name="lineEdit_4">
+     <property name="enabled">
+      <bool>false</bool>
+     </property>
+    </widget>
+   </item>
+   <item row="0" column="1">
+    <widget class="QLineEdit" name="lineEdit_2"/>
+   </item>
+   <item row="2" column="0">
+    <widget class="QLabel" name="label_4">
+     <property name="text">
+      <string>旁白语速：</string>
+     </property>
+    </widget>
+   </item>
+   <item row="3" column="1" alignment="Qt::AlignHCenter">
+    <widget class="QDialogButtonBox" name="buttonBox">
      <property name="toolTip">
       <string/>
      </property>
@@ -38,130 +63,22 @@
       <bool>false</bool>
      </property>
     </widget>
-  <widget class="QLabel" name="label">
-   <property name="geometry">
-    <rect>
-     <x>40</x>
-     <y>40</y>
-     <width>72</width>
-     <height>15</height>
-    </rect>
-   </property>
-   <property name="text">
-    <string>视频路径:</string>
-   </property>
-  </widget>
-  <widget class="QLineEdit" name="lineEdit">
-   <property name="geometry">
-    <rect>
-     <x>130</x>
-     <y>40</y>
-     <width>231</width>
-     <height>21</height>
-    </rect>
-   </property>
-  </widget>
-  <widget class="QPushButton" name="pushButton">
-   <property name="geometry">
-    <rect>
-     <x>390</x>
-     <y>40</y>
-     <width>101</width>
-     <height>31</height>
-    </rect>
-   </property>
-   <property name="text">
-    <string>选择文件</string>
-   </property>
-  </widget>
+   </item>
+   <item row="0" column="0">
     <widget class="QLabel" name="label_2">
-   <property name="geometry">
-    <rect>
-     <x>20</x>
-     <y>90</y>
-     <width>111</width>
-     <height>16</height>
-    </rect>
-   </property>
      <property name="text">
       <string>旁白脚本表格：</string>
      </property>
     </widget>
-  <widget class="QLineEdit" name="lineEdit_2">
-   <property name="geometry">
-    <rect>
-     <x>130</x>
-     <y>90</y>
-     <width>231</width>
-     <height>21</height>
-    </rect>
-   </property>
-  </widget>
-  <widget class="QPushButton" name="pushButton_2">
-   <property name="geometry">
-    <rect>
-     <x>390</x>
-     <y>90</y>
-     <width>101</width>
-     <height>31</height>
-    </rect>
-   </property>
-   <property name="text">
-    <string>表格路径</string>
-   </property>
-  </widget>
+   </item>
+   <item row="1" column="0">
     <widget class="QLabel" name="label_3">
-   <property name="geometry">
-    <rect>
-     <x>30</x>
-     <y>140</y>
-     <width>111</width>
-     <height>16</height>
-    </rect>
-   </property>
      <property name="text">
       <string>旁白说话人：</string>
      </property>
     </widget>
-  <widget class="QLabel" name="label_4">
-   <property name="geometry">
-    <rect>
-     <x>30</x>
-     <y>200</y>
-     <width>111</width>
-     <height>16</height>
-    </rect>
-   </property>
-   <property name="text">
-    <string>旁白语速：</string>
-   </property>
-  </widget>
-  <widget class="QLineEdit" name="lineEdit_3">
-   <property name="enabled">
-    <bool>false</bool>
-   </property>
-   <property name="geometry">
-    <rect>
-     <x>130</x>
-     <y>140</y>
-     <width>231</width>
-     <height>21</height>
-    </rect>
-   </property>
-  </widget>
-  <widget class="QLineEdit" name="lineEdit_4">
-   <property name="enabled">
-    <bool>false</bool>
-   </property>
-   <property name="geometry">
-    <rect>
-     <x>130</x>
-     <y>200</y>
-     <width>231</width>
-     <height>21</height>
-    </rect>
-   </property>
-  </widget>
+   </item>
+  </layout>
  </widget>
  <resources/>
  <connections>

assemble_dialog_ui.py

 # -*- coding: utf-8 -*-
 
-# Form implementation generated from reading ui file 'assemble_dialog.ui'
+# Form implementation generated from reading ui file 'd:\AddCaption\cur_version\accessibility_movie_2\assemble_dialog.ui'
 #
-# Created by: PyQt5 UI code generator 5.12
+# Created by: PyQt5 UI code generator 5.15.4
 #
-# WARNING! All changes made in this file will be lost!
+# WARNING: Any manual changes made to this file will be lost when pyuic5 is
+# run again.  Do not edit this file unless you know what you are doing.
+
 
 from PyQt5 import QtCore, QtGui, QtWidgets
 
@@ -13,46 +15,48 @@ class Ui_Dialog(object):
     def setupUi(self, Dialog):
         Dialog.setObjectName("Dialog")
         Dialog.resize(585, 326)
-        self.buttonBox = QtWidgets.QDialogButtonBox(Dialog)
-        self.buttonBox.setGeometry(QtCore.QRect(220, 250, 341, 32))
-        self.buttonBox.setToolTip("")
-        self.buttonBox.setToolTipDuration(-1)
-        self.buttonBox.setOrientation(QtCore.Qt.Horizontal)
-        self.buttonBox.setStandardButtons(QtWidgets.QDialogButtonBox.Cancel|QtWidgets.QDialogButtonBox.Ok)
-        self.buttonBox.setCenterButtons(False)
-        self.buttonBox.setObjectName("buttonBox")
-        self.label = QtWidgets.QLabel(Dialog)
-        self.label.setGeometry(QtCore.QRect(40, 40, 72, 15))
-        self.label.setObjectName("label")
-        self.lineEdit = QtWidgets.QLineEdit(Dialog)
-        self.lineEdit.setGeometry(QtCore.QRect(130, 40, 231, 21))
-        self.lineEdit.setObjectName("lineEdit")
+        self.gridLayout = QtWidgets.QGridLayout(Dialog)
+        self.gridLayout.setObjectName("gridLayout")
         self.pushButton = QtWidgets.QPushButton(Dialog)
-        self.pushButton.setGeometry(QtCore.QRect(390, 40, 101, 31))
         self.pushButton.setObjectName("pushButton")
-        self.label_2 = QtWidgets.QLabel(Dialog)
-        self.label_2.setGeometry(QtCore.QRect(20, 90, 111, 16))
-        self.label_2.setObjectName("label_2")
-        self.lineEdit_2 = QtWidgets.QLineEdit(Dialog)
-        self.lineEdit_2.setGeometry(QtCore.QRect(130, 90, 231, 21))
-        self.lineEdit_2.setObjectName("lineEdit_2")
-        self.pushButton_2 = QtWidgets.QPushButton(Dialog)
-        self.pushButton_2.setGeometry(QtCore.QRect(390, 90, 101, 31))
-        self.pushButton_2.setObjectName("pushButton_2")
-        self.label_3 = QtWidgets.QLabel(Dialog)
-        self.label_3.setGeometry(QtCore.QRect(30, 140, 111, 16))
-        self.label_3.setObjectName("label_3")
+        self.gridLayout.addWidget(self.pushButton, 0, 2, 1, 1)
         self.label_4 = QtWidgets.QLabel(Dialog)
-        self.label_4.setGeometry(QtCore.QRect(30, 200, 111, 16))
         self.label_4.setObjectName("label_4")
+        self.gridLayout.addWidget(self.label_4, 3, 0, 1, 1)
+        self.label_3 = QtWidgets.QLabel(Dialog)
+        self.label_3.setObjectName("label_3")
+        self.gridLayout.addWidget(self.label_3, 2, 0, 1, 1)
         self.lineEdit_3 = QtWidgets.QLineEdit(Dialog)
         self.lineEdit_3.setEnabled(False)
-        self.lineEdit_3.setGeometry(QtCore.QRect(130, 140, 231, 21))
         self.lineEdit_3.setObjectName("lineEdit_3")
+        self.gridLayout.addWidget(self.lineEdit_3, 2, 1, 1, 1)
         self.lineEdit_4 = QtWidgets.QLineEdit(Dialog)
         self.lineEdit_4.setEnabled(False)
-        self.lineEdit_4.setGeometry(QtCore.QRect(130, 200, 231, 21))
         self.lineEdit_4.setObjectName("lineEdit_4")
+        self.gridLayout.addWidget(self.lineEdit_4, 3, 1, 1, 1)
+        self.lineEdit = QtWidgets.QLineEdit(Dialog)
+        self.lineEdit.setObjectName("lineEdit")
+        self.gridLayout.addWidget(self.lineEdit, 0, 1, 1, 1)
+        self.label_2 = QtWidgets.QLabel(Dialog)
+        self.label_2.setObjectName("label_2")
+        self.gridLayout.addWidget(self.label_2, 1, 0, 1, 1)
+        self.lineEdit_2 = QtWidgets.QLineEdit(Dialog)
+        self.lineEdit_2.setObjectName("lineEdit_2")
+        self.gridLayout.addWidget(self.lineEdit_2, 1, 1, 1, 1)
+        self.label = QtWidgets.QLabel(Dialog)
+        self.label.setObjectName("label")
+        self.gridLayout.addWidget(self.label, 0, 0, 1, 1)
+        self.pushButton_2 = QtWidgets.QPushButton(Dialog)
+        self.pushButton_2.setObjectName("pushButton_2")
+        self.gridLayout.addWidget(self.pushButton_2, 1, 2, 1, 1)
+        self.buttonBox = QtWidgets.QDialogButtonBox(Dialog)
+        self.buttonBox.setToolTip("")
+        self.buttonBox.setToolTipDuration(-1)
+        self.buttonBox.setOrientation(QtCore.Qt.Horizontal)
+        self.buttonBox.setStandardButtons(QtWidgets.QDialogButtonBox.Cancel|QtWidgets.QDialogButtonBox.Ok)
+        self.buttonBox.setCenterButtons(False)
+        self.buttonBox.setObjectName("buttonBox")
+        self.gridLayout.addWidget(self.buttonBox, 4, 1, 1, 1, QtCore.Qt.AlignHCenter)
 
         self.retranslateUi(Dialog)
         self.buttonBox.accepted.connect(Dialog.accept)
@@ -62,9 +66,9 @@ class Ui_Dialog(object):
     def retranslateUi(self, Dialog):
         _translate = QtCore.QCoreApplication.translate
         Dialog.setWindowTitle(_translate("Dialog", "Dialog"))
-        self.label.setText(_translate("Dialog", "视频路径:"))
         self.pushButton.setText(_translate("Dialog", "选择文件"))
+        self.label_4.setText(_translate("Dialog", "旁白语速："))
+        self.label_3.setText(_translate("Dialog", "旁白说话人："))
         self.label_2.setText(_translate("Dialog", "旁白脚本表格："))
+        self.label.setText(_translate("Dialog", "视频路径:"))
         self.pushButton_2.setText(_translate("Dialog", "表格路径"))
-        self.label_3.setText(_translate("Dialog", "旁白说话人："))
-        self.label_4.setText(_translate("Dialog", "旁白语速："))
\ No newline at end of file

detect_with_ocr.py

@@ -186,7 +186,9 @@ def detect_subtitle(img: np.ndarray) -> Union[str, None]:
     subTitle = ''
     height = down_b - up_b
     img = img[int(up_b - height * 0.7):int(down_b + height * 0.7)]
-    # img = cv2.resize(img, (int(img.shape[1] * 0.5), int(img.shape[0] * 0.5)))
+    # 针对低帧率的视频做图像放大处理
+    if img.shape[1] < 1000:
+        img = cv2.resize(img, (int(img.shape[1] * 1.5), int(img.shape[0] * 1.5)))
     res = ocr.ocr(img, cls=True)
     sorted(res, key=lambda text: text[0][0][1])
     if len(res) == 0:

main_window.py

@@ -655,7 +655,7 @@ class MainWindow(QMainWindow, Ui_MainWindow):
         previewed_audio[os.path.basename(path)] = 1
 
     def openVideoFile(self):
-        path = QFileDialog.getOpenFileUrl(self, "选择待导入视频", QUrl(os.getcwd()), "Video Files(*.mp4 *.rmvb *mkv *avi)")[0]
+        path = QFileDialog.getOpenFileUrl(self, "选择待导入视频", QUrl(os.getcwd()), "Video Files(*.mp4 *.rmvb *mkv *avi *.);; 所有文件(*.*)")[0]
         print(path.url()[8:])
         if len(path.url()) == 0 or not os.path.exists(path.url()[8:]):
             return path
@@ -801,10 +801,6 @@ class MainWindow(QMainWindow, Ui_MainWindow):
 
 
     def btn_pressed_preview_audio_slot(self):
-        for t in self.all_threads:
-            if t.name == "single_speech_synthesis" and t.is_alive():
-                self.prompt_dialog.show_with_msg("音频正在合成，请稍候")
-                return 
         btn = self.sender()
         # 法1：按照物理位置。这样的结果不太对
         idx = self.pb_tableWidget.indexAt(btn.pos())
@@ -816,6 +812,13 @@ class MainWindow(QMainWindow, Ui_MainWindow):
         audio_path = os.path.dirname(self.projectContext.excel_path) + (
                         "/tmp/%.2f.wav" % pos_sec)
         print("待播放的音频文件为", audio_path)
+        if audio_path is not None:
+        # 确认该音频是否正在合成中
+            for t in self.all_threads:
+                if t.name == "single_speech_synthesis" and t.is_alive():
+                    if audio_path in t._args:
+                        self.prompt_dialog.show_with_msg("音频正在合成，请稍候")
+                        return 
         # 2、如果找到了该音频，则新起一个线程播放
         if audio_path != None and os.path.exists(audio_path):
             t = RunThread(funcName=self.play_audio,
@@ -1325,9 +1328,14 @@ class MainWindow(QMainWindow, Ui_MainWindow):
     def change_audio_speed(self):
         combo = self.sender()
         idx = self.pb_tableWidget.indexAt(combo.pos())
-        print("index:", idx.row())
+        row = idx.row()
+        print("index:", row)
         # 将audio_player的资源置空
         print("释放当前播放的音频文件")
         self.audio_player.setMedia(QMediaContent())
-        self.projectContext.aside_list[idx.row()].speed = combo.currentText()
-        self.do_generate_audio_by_aside_row(idx.row())
+        self.projectContext.aside_list[row].speed = combo.currentText()
+        # 更新字幕旁白表格里对应行的语速
\ No newline at end of file
+        all_idx = self.projectContext.aside_subtitle_2contentId(self.projectContext.aside_list[row])
+        self.projectContext.all_elements[int(all_idx)].speed = combo.currentText()
+        self.all_tableWidget.setItem(int(all_idx), constant.Content.SpeedColumnNumber, QTableWidgetItem(combo.currentText()))
+        self.do_generate_audio_by_aside_row(row)
\ No newline at end of file

main_window_ui.py

@@ -21,6 +21,7 @@ class Ui_MainWindow(object):
 "    padding: 5px;\n"
 "}")
         self.centralwidget = QtWidgets.QWidget(MainWindow)
+        # self.centralwidget.setStyleSheet("font: 9pt \"黑体\";")
         self.centralwidget.setObjectName("centralwidget")
         self.horizontalLayout = QtWidgets.QHBoxLayout(self.centralwidget)
         self.horizontalLayout.setObjectName("horizontalLayout")
@@ -206,6 +207,7 @@ class Ui_MainWindow(object):
         self.horizontalLayout_4 = QtWidgets.QHBoxLayout(self.all_tab)
         self.horizontalLayout_4.setObjectName("horizontalLayout_4")
         self.all_tableWidget = QtWidgets.QTableWidget(self.all_tab)
+        # self.all_tableWidget.setStyleSheet("font: 9pt \"微软雅黑\";")
         self.all_tableWidget.setObjectName("all_tableWidget")
         self.all_tableWidget.setColumnCount(0)
         self.all_tableWidget.setRowCount(0)
@@ -240,6 +242,7 @@ class Ui_MainWindow(object):
         self.shuiping.setStretch(1, 5)
         self.verticalLayout.addLayout(self.shuiping)
         self.verticalWidget_2 = QtWidgets.QWidget(self.centralwidget)
+        # self.verticalWidget_2.setStyleSheet("font: 9pt \"微软雅黑\";")
         self.verticalWidget_2.setObjectName("verticalWidget_2")
         self.chuizhi = QtWidgets.QVBoxLayout(self.verticalWidget_2)
         self.chuizhi.setObjectName("chuizhi")
@@ -330,18 +333,26 @@ class Ui_MainWindow(object):
         self.statusbar.setObjectName("statusbar")
         MainWindow.setStatusBar(self.statusbar)
         self.setting = QtWidgets.QAction(MainWindow)
+        # font = QtGui.QFont()
+        # font.setFamily("微软雅黑")
+        # self.setting.setFont(font)
         self.setting.setObjectName("setting")
         self.action_open_project = QtWidgets.QAction(MainWindow)
+        # self.action_open_project.setFont(font)
         self.action_open_project.setObjectName("action_open_project")
         self.import_movie = QtWidgets.QAction(MainWindow)
+        # self.import_movie.setFont(font)
         self.import_movie.setObjectName("import_movie")
         self.actions = QtWidgets.QAction(MainWindow)
         self.actions.setObjectName("actions")
         self.action_save = QtWidgets.QAction(MainWindow)
+        # self.action_save.setFont(font)
         self.action_save.setObjectName("action_save")
         self.action_undo = QtWidgets.QAction(MainWindow)
+        # self.action_undo.setFont(font)
         self.action_undo.setObjectName("action_undo")
         self.action_redo = QtWidgets.QAction(MainWindow)
+        # self.action_redo.setFont(font)
         self.action_redo.setObjectName("action_redo")
         self.action_3 = QtWidgets.QAction(MainWindow)
         self.action_3.setObjectName("action_3")
@@ -352,10 +363,12 @@ class Ui_MainWindow(object):
         self.action_operate = QtWidgets.QAction(MainWindow)
         self.action_operate.setObjectName("action_operate")
         self.action_export = QtWidgets.QAction(MainWindow)
+        # self.action_export.setFont(font)
         self.action_export.setObjectName("action_export")
         self.action_insert_aside_from_now = QtWidgets.QAction(MainWindow)
         self.action_insert_aside_from_now.setObjectName("action_insert_aside_from_now")
         self.action_create = QtWidgets.QAction(MainWindow)
+        # self.action_create.setFont(font)
         self.action_create.setObjectName("action_create")
         self.menu.addAction(self.action_create)
         self.menu.addAction(self.action_open_project)

management.py

@@ -174,6 +174,7 @@ class ProjectContext:
             return 
         with open(self.conf_path, 'r', encoding='utf8') as f:
             info = json.load(f)
+            # print(json.dumps(info, ensure_ascii=False, indent=4))
             self.video_path = info["video_path"]
             self.excel_path = info["excel_path"]
             self.speaker_info = info["speaker_info"]["speaker_id"]
@@ -208,7 +209,6 @@ class ProjectContext:
             }
             f.write(json.dumps(info))
  
-
     def setVideoPath(self, video_path):
         self.video_path = video_path
 

operation_dialog.py

@@ -141,7 +141,7 @@ class Operation_Dialog(QDialog, Ui_Dialog):
                     assert start_time_f < end_time_f
                     if row_number < rowCount:
                         # 要求结束时间在下一行的起始时间前面，对增加和修改都适用
-                        assert float(self.mainWindow.projectContext.all_elements[list_idx+1].st_time_sec) > end_time_f
+                        assert float(self.mainWindow.projectContext.all_elements[list_idx+1].st_time_sec) >= end_time_f
 
                 # 要求起始时间在上一行后面，在下一行前面
                 if self.comboBox_2.currentText() == "增加一行":