add Simplified Chinese and Traditional Chinese convert function #77

fasterWhisperGUIConfig.json

@@ -8,9 +8,9 @@
     "model_param": {
         "localModel": true,
         "onlineModel": false,
-        "model_path": "",
+        "model_path": "F:/WhisperModels/faster-whisper/large-v3-float32",
         "modelName": 0,
-        "use_v3_model": false,
+        "use_v3_model": true,
         "device": 1,
         "deviceIndex": "0",
         "preciese": 0,
@@ -31,15 +31,15 @@
     "setting": {
         "saveConfig": true,
         "autoLoadModel": false,
-        "language": 0,
+        "language": 2,
         "huggingface_user_token": "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",
         "autoGoToOutputPage": 2,
         "autoClearTempFiles": true,
         "themeColor": "#009faa"
     },
     "Transcription_param": {
         "aggregate_contents": true,
-        "language": 3,
+        "language": 2,
         "task": false,
         "beam_size": "5",
         "best_of": "5",
@@ -67,7 +67,7 @@
         "tabMovable": false,
         "tabScrollable": false,
         "tabShadowEnabled": false,
-        "tabMaxWidth": 253,
+        "tabMaxWidth": 205,
         "closeDisplayMode": 0,
         "whisperXMinSpeaker": 3,
         "whisperXMaxSpeaker": 3,

faster_whisper_GUI/mainWindows.py

@@ -32,6 +32,8 @@
                         )
 
 from faster_whisper.transcribe import TranscriptionInfo
+from faster_whisper import Word
+
 import torch
 
 from .config import (
@@ -60,6 +62,9 @@
 from .util import outputWithDateTime
 from .split_audio import SplitAudioFileWithSpeakersWorker
 
+import opencc
+
+
 # =======================================================================================
 # SignalStore
 # =======================================================================================
@@ -397,6 +402,7 @@ def transcribeProcess(self):
             # 转写参数
             Transcribe_params : dict = self.getParamTranscribe()
 
+            print(f"language:{Transcribe_params['language']}")
             print("Transcribes options:")
             self.log.write("Transcribes options:\n")
 
@@ -558,6 +564,44 @@ def createResultInTable(self, results):
 
         print(f"len_model: {len(self.tableModel_list)}")
 
+    def simplifiedAndTraditionalChineseConvert(self, segments, language):
+        # 設置轉換器
+                    if language == "Auto" or language == "zhs":
+                        print(f"convert to Simplified Chinese")
+                        print(f"len:{len(segments)}")
+                        cc = opencc.OpenCC('t2s')
+
+                        # for segment in segment_:
+                            # new_text = cc.convert(segment.text)
+                            # print(f"[{segment.text} --> {new_text}]")
+                            # segment.text = new_text
+                            # print(f"len_words: {len(segment.words)}")
+                            # if len(segment.words) > 0:
+                            #     for word in segment.words:
+                            #         new_word = cc.convert(word.word)
+                            #         print(f"    {word.word} --> {new_word}")
+                            #         word.word = new_word
+                    elif language == "zht":
+                        print(f"convert to Traditional Chinese")
+                        print(f"len:{len(segments)}")
+                        cc = opencc.OpenCC('s2t')
+
+                    # 轉換簡繁
+                    for segment in segments:
+
+                        new_text = cc.convert(segment.text)
+                        # print(f"[{segment.text} --> {new_text}]")
+                        segment.text = new_text
+                        # print(f"len_words: {len(segment.words)}")
+                        if len(segment.words) > 0:
+
+                            for word in segment.words:
+                                new_word = cc.convert(word.word)
+                                # print(f"    {word.word} --> {new_word}")
+
+                                word = Word(word.start,word.end,new_word,word.probability)
+                                # word.word = new_word
+
     def transcribeOver(self, segments_path_info:list):
         # self.button_process.clicked.disconnect(self.cancelTrancribe)
         # self.button_process.clicked.connect(self.onButtonProcessClicked)
@@ -582,13 +626,15 @@ def transcribeOver(self, segments_path_info:list):
                             )
 
             self.result_faster_whisper = segments_path_info
-            # for segments in self.result_faster_whisper:
-            #     segment_, path, info = segments
-            #     print(path, info.language)
-            #     print(f"len:{len(segment_)}")
-            #     for segment in segment_:
-            #         print(f"[{segment.start}s --> {segment.end}] | {segment.text}")
-            #         print(f"len_words: {len(segment.words)}")
+
+            for segments in self.result_faster_whisper:
+                segment_, path, info = segments
+                if info.language == "zh":
+                    print(path, info.language)
+                    language_param = self.page_transcribes.combox_language.currentText().split("-")[0]
+
+                    self.simplifiedAndTraditionalChineseConvert(segment_,language_param)
+
 
             print(f"len_segments_path_info_result: {len(segments_path_info)}")
             self.showResultInTable(self.result_faster_whisper)
@@ -622,7 +668,7 @@ def getParamTranscribe(self) -> dict:
             language = None
         if language in ["zht","zhs"]:
             language = "zh"
-            
+
         Transcribe_params["language"] = language
 
         task = self.page_transcribes.switchButton_Translate_to_English.isChecked()
@@ -1115,11 +1161,13 @@ def openExcitedFiles(self):
             self.raiseErrorInfoBar(self.tr("读取字幕文件失败"), self.tr("读取失败 文件可能使用了不同的编码\n检查 fasterwhispergui.log 文件可能会获取更多信息"))
             return
         # 输出字幕文件内容
-        for segment in segments:
-            print(f"[{segment.start}s --> {segment.end}s] | {segment.speaker+':'+segment.text if segment.speaker else segment.text}")
+        # for segment in segments:
+        #     print(f"[{segment.start}s --> {segment.end}s] | {segment.speaker+':'+segment.text if segment.speaker else segment.text}")
 
+        language = self.page_transcribes.combox_language.currentText().split("-")[0]
+
         info = TranscriptionInfo(
-                                    language=self.page_transcribes.combox_language.currentText().split("-")[0],
+                                    language="zh" if language in ["zhs","zht"] else language,
                                     language_probability=1,
                                     duration=None, 
                                     duration_after_vad=None,
@@ -1128,6 +1176,9 @@ def openExcitedFiles(self):
                                     vad_options={}
                                 )
 
+        if language in ["zhs", "zht"]:
+            self.simplifiedAndTraditionalChineseConvert(segments, language)
+
         if file_subtitle_fileName and file:
             self.result_whisperx_aligment = None
             self.result_whisperx_speaker_diarize = None
@@ -1334,7 +1385,7 @@ def outputAudioPartWithSpeaker(self):
         # self.page_output.outputAudioPartWithSpeakerButton.setEnabled(False)
         self.setPageOutButtonStatus()
 
-        language = self.page_transcribes.combox_language.currentText().split("-")[0]
+        language = self.page_transcribes.combox_language.currentText().split("-")[-1]
 
         output_path = self.page_output.outputGroupWidget.LineEdit_output_dir.text()
         self.splitAudioFileWithSpeakerWorker = SplitAudioFileWithSpeakersWorker(self.current_result,output_path,language ,self)

faster_whisper_GUI/split_audio.py

@@ -69,7 +69,7 @@ def run(self):
             # 数据标注文件
             list_file = open(f"{output_path + '/' + '00_list.csv'}","w",encoding="utf8")
             # 格式：vocal_path|speaker_name|language|text
-            list_file.write("vocal_path,speaker_name,language,text\n")
+            list_file.write("vocal_path,    speaker_name,    language,    text\n")
 
             for segment in segments:
                 # if not segment.speaker : continue
@@ -93,7 +93,7 @@ def run(self):
                 output_fileName = output_fileName.replace('\\','/')
 
                 # 输出标注信息
-                list_file.write(f"{output_fileName},{speaker},{self.language},{segment.text}\n")
+                list_file.write(f"{output_fileName},{speaker},{self.language},{segment.text.strip().replace(',',' ')}\n")
 
         list_file.close()
         # 完成后发送结果信号

faster_whisper_GUI/transcribe.py

@@ -281,9 +281,12 @@ def transcribe_file(self, file) -> (TranscriptionInfo, List): # type: ignore
         return info, segmentsTranscribe
 
     def detect_Audio_info(self, info):
-        language = Language_dict[info.language]
-        if language:
-            language = language.capitalize()
+        if info.language != "zh":
+            language = Language_dict[info.language]
+            if language:
+                language = language.capitalize()
+        else:
+            language = "Chinese" 
         language_probability = info.language_probability
         duration = info.duration
         duration = secondsToHMS(duration).replace(",", ".")
@@ -460,7 +463,10 @@ def writeSMI(fileName:str, segments:List[segment_Transcribe], language:str, avFi
             i += 1
     else:
         smi += "  #SUB{color: white; background-color: black; font-family: Arial; font-size: 12pt; font-weight: normal; text-align: left;}"
-    smi += f"  .{language_type_CC} {'{'} name: {Language_dict[language].capitalize()}; lang: {language}; SAMIType: CC; {'}'}\n"
+    if language != "zh":
+        smi += f"  .{language_type_CC} {'{'} name: {Language_dict[language].capitalize()}; lang: {language}; SAMIType: CC; {'}'}\n"
+    else:
+        smi += f"  .{language_type_CC} {'{'} name: {'Chinese'}; lang: {language}; SAMIType: CC; {'}'}\n"
     smi += "-->\n"
     smi += "</STYLE>\n"
     smi += "</HEAD>\n"