finish function split Audio #54

fasterWhisperGUIConfig.json

@@ -3,29 +3,29 @@
     "demucs": {
         "overlap": 0.1,
         "segment": 7.8,
-        "tracks": 0
+        "tracks": 1
     },
     "model_param": {
         "localModel": true,
         "onlineModel": false,
-        "model_path": "",
+        "model_path": "F:/WhisperModels/faster-whisper/large-v3-float32",
         "modelName": 0,
         "use_v3_model": true,
         "device": 1,
         "deviceIndex": "0",
         "preciese": 0,
         "thread_num": "4",
         "num_worker": "1",
-        "download_root": "",
+        "download_root": "C:/Users/12059/.cache/huggingface/hub",
         "local_files_only": false
     },
     "vad_param": {
         "use_VAD": true,
-        "threshold": 0.3,
+        "threshold": 0.5,
         "minSpeechDuration": "250",
         "minSilenceDuration": "2000",
         "maxSpeechDuration": "inf",
-        "windowSize": 2,
+        "windowSize": 1,
         "speechPad": "400"
     },
     "setting": {
@@ -37,9 +37,9 @@
         "autoClearTempFiles": true
     },
     "Transcription_param": {
-        "language": 8,
+        "language": 2,
         "task": false,
-        "beam_size": "5",
+        "beam_size": "2",
         "best_of": "1",
         "patience": "1.0",
         "length_penalty": "1.0",
@@ -65,10 +65,10 @@
         "tabMovable": false,
         "tabScrollable": true,
         "tabShadowEnabled": false,
-        "tabMaxWidth": 326,
+        "tabMaxWidth": 300,
         "closeDisplayMode": 0,
-        "whisperXMinSpeaker": 3,
-        "whisperXMaxSpeaker": 3,
+        "whisperXMinSpeaker": 2,
+        "whisperXMaxSpeaker": 2,
         "outputFormat": 0,
         "outputEncoding": 1
     }

faster_whisper_GUI/UI_MainWindows.py

@@ -190,6 +190,13 @@ def __init__(self, translator:QTranslator=None):
         # 设置配置
         self.setConfig()
 
+        # 根据读取的配置设置完控件状态之后，根据控件状态设置相关属性
+        self.page_output.tableTab.onDisplayModeChanged(self.page_output.tableTab.closeDisplayModeComboBox.currentIndex())
+        self.page_output.tableTab.tabBar.setMovable(self.page_output.tableTab.movableCheckBox.isChecked())
+        self.page_output.tableTab.tabBar.setScrollable(self.page_output.tableTab.scrollableCheckBox.isChecked())
+        self.page_output.tableTab.tabBar.setTabShadowEnabled(self.page_output.tableTab.shadowEnabledCheckBox.isChecked())
+        self.page_output.tableTab.tabBar.setTabMaximumWidth(self.page_output.tableTab.tabMaxWidthSpinBox.value())
+
     def initWin(self):
 
         self.setObjectName("FramlessMainWin")

faster_whisper_GUI/mainWindows.py

@@ -882,20 +882,23 @@ def aligmentOver(self, segments_path_info:list):
         self.whisperXWorker = None
 
     def whisperXAligmentTimeStample(self):
-        if self.result_faster_whisper is None :
+        if self.result_faster_whisper is None and self.current_result is None:
             self.raiseErrorInfoBar(
                                     self.tr("错误"),
                                     self.tr("没有有效的 音频-字幕 转写结果，无法进行对齐")
             )
             return
+
+        elif self.current_result is None :
+            self.current_result = self.result_faster_whisper
 
         self.setPageOutButtonStatus()
         self.outputWithDateTime("TimeStample_Alignment")
 
         self.setStateTool(title=self.tr("WhisperX"), text=self.tr("时间戳对齐"), status=False)
 
         if self.whisperXWorker is None:
-            self.whisperXWorker = WhisperXWorker(self.result_faster_whisper, alignment=True,speaker_diarize=False, parent=self)
+            self.whisperXWorker = WhisperXWorker(self.current_result, alignment=True,speaker_diarize=False, parent=self)
         else:
             self.whisperXWorker.result_segments_path_info = self.result_faster_whisper
             self.whisperXWorker.alignment = True
@@ -909,7 +912,7 @@ def whisperXDiarizeSpeakers(self):
 
         whisperParams = self.getParamWhisperX()
 
-        result_needed = self.result_whisperx_aligment or self.result_faster_whisper
+        result_needed = self.current_result or self.result_whisperx_aligment or self.result_faster_whisper
         # print(f"result_useing: {result_needed}")
         # try:
         #     print(len(result_needed))
@@ -956,6 +959,7 @@ def setPageOutButtonStatus(self):
         self.page_output.WhisperXAligmentTimeStampleButton.setEnabled(not self.page_output.WhisperXAligmentTimeStampleButton.isEnabled())
         self.page_output.outputSubtitleFileButton.setEnabled(not self.page_output.outputSubtitleFileButton.isEnabled())
         self.page_output.WhisperXSpeakerDiarizeButton.setEnabled(not self.page_output.WhisperXSpeakerDiarizeButton.isEnabled())
+        self.page_output.outputAudioPartWithSpeakerButton.setEnabled(not self.page_output.outputAudioPartWithSpeakerButton.isEnabled()) 
 
     def speakerDiarizeOver(self, segments_path_info:list):
         self.setPageOutButtonStatus()
@@ -1004,7 +1008,7 @@ def is_audio_or_video(self, file_path:str) -> bool:
 
         # 判定打开的文件是否音视频文件
         try:
-            av_cont = av.open(file_path)
+            av_cont = av.open(file_path, metadata_errors = "ignore")
 
             # 获取文件的全部 流数据
             av_streams = av_cont.streams
@@ -1105,12 +1109,13 @@ def openExcitedFiles(self):
             self.result_whisperx_aligment = None
             self.result_whisperx_speaker_diarize = None
 
-            if self.result_faster_whisper is not None:
-                self.result_faster_whisper.append((segments, file, info))
+            if self.current_result is not None:
+                self.current_result.append((segments, file, info))
             else:
-                self.result_faster_whisper = [(segments, file, info)]
+                self.current_result = [(segments, file, info)]
             # self.tableModel_list[file] = file_subtitle_fileName
-            self.showResultInTable(self.result_faster_whisper)
+
+            self.showResultInTable(self.current_result)
 
     def reSetButton_demucs_process(self):
         self.page_demucs.process_button.setText(self.tr("提取"))
@@ -1287,16 +1292,26 @@ def outputAudioPartWithSpeaker(self):
         """
         outputWithDateTime("SegmentAudioFileWithSpeaker")
 
-        self.page_output.outputAudioPartWithSpeakerButton.setEnabled(False)
+        # self.page_output.outputAudioPartWithSpeakerButton.setEnabled(False)
+        self.setPageOutButtonStatus()
+
         output_path = self.page_output.outputGroupWidget.LineEdit_output_dir.text()
         self.splitAudioFileWithSpeakerWorker = SplitAudioFileWithSpeakersWorker(self.current_result,output_path, self)
         self.splitAudioFileWithSpeakerWorker.result_signal.connect(self.splitAudioFileWithSpeakerWorkerFinished)
+        self.splitAudioFileWithSpeakerWorker.current_task_signal.connect(lambda file: self.setStateTool(self.tr("分割音频"), self.tr("处理文件：") + file, False))
         self.splitAudioFileWithSpeakerWorker.start()
+
+        self.setStateTool(self.tr("分割音频"), self.tr("按说话人分割音频文件"), False)
 
     def splitAudioFileWithSpeakerWorkerFinished(self):
-        mes = MessageBox("over","ok", self)
-        mes.show()
-        self.page_output.outputAudioPartWithSpeakerButton.setEnabled(True)
+
+        self.setStateTool(self.tr("分割音频"), self.tr("按说话人分割音频文件完成"), True)
+        self.raiseSuccessInfoBar(self.tr("分割音频完成"),self.tr("按说话人分割音频文件完成"))
+
+        # mes = MessageBox("over","ok", self)
+        # mes.show()
+        # self.page_output.outputAudioPartWithSpeakerButton.setEnabled(True)
+        self.setPageOutButtonStatus()
 
     def singleAndSlotProcess(self):
         """

faster_whisper_GUI/modelPageNavigationInterface.py

@@ -87,6 +87,7 @@ def setupUI(self):
         self.label_model_path.setObjectName("LabelModelPath")
         self.label_model_path.setStyleSheet("#LabelModelPath{ background : rgba(0, 128, 0, 120); }")
         self.lineEdit_model_path = LineEdit()
+        self.lineEdit_model_path.setClearButtonEnabled(True)
         self.toolPushButton_get_model_path = ToolButton()
         self.toolPushButton_get_model_path.setIcon(self.style().standardPixmap(QStyle.StandardPixmap.SP_DirOpenIcon))
 

faster_whisper_GUI/outputPageNavigationInterface.py

@@ -71,7 +71,7 @@ def setupUI(self):
 
         self.outputAudioPartWithSpeakerButton = PushButton()
         self.outputAudioPartWithSpeakerButton.setText(self.tr("输出音频分段"))
-        self.outputAudioPartWithSpeakerButton.setToolTip(self.tr("将音频按照说话人进行分段，并输出"))
+        self.outputAudioPartWithSpeakerButton.setToolTip(self.tr("将音频按照说话人和时间戳进行分段处理，并输出"))
 
         self.WhisperXHBoxLayout = QHBoxLayout()
         self.WhisperXHBoxLayout.addWidget(self.WhisperXAligmentTimeStampleButton, 0, Qt.AlignmentFlag.AlignLeft)

faster_whisper_GUI/settingPageNavigation.py

@@ -13,7 +13,8 @@
                             ComboBox, 
                             LineEdit,
                             MessageBox,
-                            PushButton
+                            PushButton,
+                            TitleLabel
                         )
 
 from .paramItemWidget import ParamWidget
@@ -52,6 +53,12 @@ def addLayout(self, layout):
         self.mainLayout.addLayout(layout)
 
     def setupUI(self):
+
+        self.mainLayout.addSpacing(10)
+
+        self.titleLabel_title = TitleLabel(self.__tr("软件设置"))
+        self.addWidget(self.titleLabel_title)
+
         # --------------------------------------------------------------------------------------------------------------------------------------------------------------
         self.switchButton_saveConfig = SwitchButton()
         self.switchButton_saveConfig.setChecked(True)

faster_whisper_GUI/split_audio.py

@@ -6,6 +6,7 @@
 class SplitAudioFileWithSpeakersWorker(QThread):
     # 定义一个信号，用于在处理完成后发送结果
     result_signal = Signal(str)
+    current_task_signal = Signal(str)
 
     def __init__(self, segments_path_info_list:list, output_path, parent=None):
         super().__init__(parent)
@@ -32,7 +33,13 @@ def creatCommandLine(self, start_time, end_time, fileName, output_path, speaker)
         return commandLine
 
     def getOutPutFileName(self, output_path:str, start_time:str, end_time:str, speaker:str):
-        return os.path.join(output_path, f"{speaker}_{start_time.replace(':','_')}_{end_time.replace(':','_')}.wav")
+        fileName = ""
+        if not(speaker is None) and speaker != "":
+            fileName = os.path.join(output_path, f"{speaker}_{start_time.replace(':','_')}_{end_time.replace(':','_')}.wav")
+        else:
+            fileName = os.path.join(output_path, f"UnClassedSpeaker_{start_time.replace(':','_')}_{end_time.replace(':','_')}.wav")
+        return fileName
+
 
     def run(self):
         self.is_running = True
@@ -42,6 +49,8 @@ def run(self):
             base_path,file = os.path.split(path)
             print(f"    current task: {file}")
 
+            self.current_task_signal.emit(file)
+
             if not self.output_path:
                 output_path = base_path
             else:
@@ -54,7 +63,8 @@ def run(self):
                 os.makedirs(output_path)
 
             for segment in segments:
-                if not segment.speaker: continue
+                # if not segment.speaker : continue
+
                 start_time = secondsToHMS(segment.start)
                 end_time = secondsToHMS(segment.end)
                 speaker = segment.speaker

faster_whisper_GUI/tableModel_segments_path_info.py

@@ -2,8 +2,11 @@
 from PySide6.QtCore import QAbstractTableModel, Qt
 # from PySide6.QtWidgets import QStyledItemDelegate, QLineEdit
 from typing import List
+from faster_whisper import  Word
 
 from .seg_ment import segment_Transcribe
+from .transcribe import secondsToHMS
+from .util import HMSToSeconds
 
 # 自定义数据模型，用于在表格中显示数据
 class TableModel(QAbstractTableModel):
@@ -22,18 +25,19 @@ def data(self, index, role):
             row, column = index.row(), index.column()
             data:segment_Transcribe = self._data[row]
             if column == 0:
-                value = data.start
+                value = secondsToHMS(data.start)
             elif column == 1:
-                value = data.end
+                value = secondsToHMS(data.end)
             elif column == 2:
                 try:
                     value = f"{data.speaker}:{data.text}" if data.speaker is not None else data.text
                 except AttributeError:
                     value = data.text
 
-            elif column == 3:    
+            elif column == 3:
                 value = ";".join([f"<{word.start}>{word.word}<{word.end}>" for word in data.words]) if len(data.words) > 0 else ""
             # print(type(value))
+            value.encode('utf-8').decode('utf-8')
             return value
 
     # 使数据可编辑
@@ -44,21 +48,51 @@ def setData(self, index, value, role):
 
             try:
                 if column == 0:
-                    self._data[row].start = float(value)
+                    self._data[row].start = HMSToSeconds(value)
                 elif column == 1:
-                    self._data[row].end = float(value)
+                    self._data[row].end = HMSToSeconds(value)
                 elif column == 2:
                     if value != "":
                         retxt = value.split(":")
                         if len(retxt) > 1:
+                            if self._data[row].speaker != retxt[0]:
+                                temp_data_speaker = self._data[row].speaker 
+                                for data in self._data:
+                                    if data.speaker is not None and data.speaker != "" and data.speaker == temp_data_speaker:
+                                        data.speaker = retxt[0]
                             self._data[row].speaker = retxt[0]
                             self._data[row].text = ":".join(retxt[1:])
                         else:
                             self._data[row].text = value
                     else:
                         return False
+
                 elif column == 3:
-                    return False
+                    if value != "":
+                        words_list = []
+                        text = ""
+                        try:
+                            words = value.split(";")
+                            print(words)
+                            for word in words:
+                                temp = word.split("<")
+                                print(temp)
+                                end_time = float(temp[-1].split(">")[0])
+                                start_time = float(temp[1].split(">")[0])
+                                word_text = temp[1].split(">")[-1]
+                                text += word_text
+                                word_ = Word(start=start_time, end=end_time, word=word_text, probability=1.0)
+                                words_list.append(word_)
+
+                            print(text)
+                            self._data[row].words = words_list
+                            self._data[row].text = text
+                        except Exception as e:
+                            print(f"edit words-level timestample error:{e}")
+                            return False
+                    else:
+                        return False
+
                 self.dataChanged.emit(index, index)
                 return True
             except ValueError:

faster_whisper_GUI/tableViewInterface.py

@@ -190,8 +190,8 @@ def addSubInterface(self
         widget.resizeColumnToContents(2)
 
         # 设置列宽
-        widget.setColumnWidth(0,100)
-        widget.setColumnWidth(1,100)
+        widget.setColumnWidth(0,110)
+        widget.setColumnWidth(1,110)
         # widget.setColumnWidth(3,500)
         # widget.setColumnWidth(4,50)