合成音频使用用户选中的语速

assemble_dialog.py

@@ -65,12 +65,13 @@ class Assemble_Dialog(QDialog, Ui_Dialog):
         print("start_assemble")
         video_path = self.lineEdit.text()
         # 默认 输出的音频是工程目录+/output
-        audio_dir = self.projectContext.project_base_dir+"output/"
+        audio_dir = os.path.join(self.projectContext.project_base_dir, "output")
         sheet_path = self.lineEdit_2.text()
         speaker_info = self.lineEdit_3.text()
         speed_info =  self.lineEdit_4.text()
         #todo 后续变成常量存起来，或者做成配置
-        caption_path = replace_path_suffix(self.lineEdit.text(), ".srt")
+        # caption_path = replace_path_suffix(self.lineEdit.text(), ".srt")
+        caption_path = os.path.join(audio_dir, os.path.basename(video_path).split('.')[0] + ".srt")
 
         print("video_path: ",video_path)
         print("audio_dir: ",audio_dir)
@@ -78,7 +79,7 @@ class Assemble_Dialog(QDialog, Ui_Dialog):
         print("speed_info: ",speed_info)
         print("caption_path: ",caption_path)
         print("speaker_info: ",speaker_info)
-        self.start_assemble_signal.emit([video_path,audio_dir, sheet_path,speed_info, caption_path, speaker_info])
+        self.start_assemble_signal.emit([video_path, audio_dir, sheet_path,speed_info, caption_path, speaker_info])
 
 if __name__ == '__main__':
     app = QApplication(sys.argv)

main_window.py

@@ -508,16 +508,15 @@ class MainWindow(QMainWindow, Ui_MainWindow):
                 self.export_timer.stop()
 
             print("===已有线程结束了 in %s ===" % (type))
-            self.statusbarLabel.setText("  %s完成" % (type))
-            self.progressBar.setValue(100)
-            self.progressLabel.setText(f"100%")
-            self.projectContext.nd_process = 1
             for t in self.threads:
                 if t.exitcode != 0:
                     print("Exception in", t.getName())
                     self.show_warning_msg_box("运行出错，请联系开发者处理")
-                    print("当前已有的检测结果", self.projectContext.all_elements)
                     return
+            self.statusbarLabel.setText("  %s完成" % (type))
+            self.progressBar.setValue(100)
+            self.progressLabel.setText(f"100%")
+            self.projectContext.nd_process = 1
 
     def deal_synthesis_callback_slot(self, threads, state):
         self.statusbarLabel.setText("  准备合成：")
@@ -1165,8 +1164,8 @@ class MainWindow(QMainWindow, Ui_MainWindow):
         self.projectContext.save_project(False)
 
     def export_all(self):
-        # 暂时存放音频的文件夹被命名为tmp
-        output_dir = os.path.join(self.projectContext.project_base_dir, "tmp")
+        # 存放合成音频的文件夹被命名为output
+        output_dir = os.path.join(self.projectContext.project_base_dir, "output")
         if os.path.exists(output_dir) and len(os.listdir(output_dir)) > 0:
             self.export.export_slot(self.projectContext.video_path, output_dir)
         else:

speech_synthesis.py

@@ -89,7 +89,6 @@ def speech_synthesis(text: str, output_file: str, speaker: Speaker, speed: float
         speaker (Speaker): 说话人
         speed (float, optional): 指定的音频语速. Defaults to 1.0.
     """
-    audio_path = tmp_file
     speech_config = SpeechConfig(
         subscription="db34d38d2d3447d482e0f977c66bd624",
         region="eastus"
@@ -102,7 +101,6 @@ def speech_synthesis(text: str, output_file: str, speaker: Speaker, speed: float
     if not os.path.exists(os.path.dirname(output_file)):  # 如果路径不存在
         print("output_file路径不存在，创建:", os.path.dirname(output_file))
         os.makedirs(os.path.dirname(output_file))
-    audio_config = AudioOutputConfig(filename=audio_path)
     synthesizer = SpeechSynthesizer(speech_config=speech_config, audio_config=None)
     ssml_string = f"""
     <speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="{speech_config.speech_synthesis_language}">
@@ -169,7 +167,7 @@ def read_sheet(book_path: str, sheet_name: str = "") -> dict:
     return sheet_content
 
 
-def get_narratage_text(sheet_content: dict, speed: float) -> Tuple[list, list, list]:
+def get_narratage_text(sheet_content: dict) -> Tuple[list, list, list]:
     """获取旁白解说文本及起止时间
 
     Args:
@@ -183,40 +181,49 @@ def get_narratage_text(sheet_content: dict, speed: float) -> Tuple[list, list, l
     subtitle = sheet_content['字幕']
     start_time = sheet_content['起始时间']
     end_time = sheet_content['终止时间']
+    speeds = sheet_content["语速"]
     narratage_start_time = []
     narratage_end_time = []
     narratage_text = []
+    narratage_speed = []
     for i, text in enumerate(narratage):
+        # 这里的speed是x.x倍速
+        speed = float(speeds[i].split('(')[0])
         if text is not None:
             if text == '翻译':
                 narratage_text.append(subtitle[i])
-                narratage_start_time.append(float(start_time[i]))
-                narratage_end_time.append(float(end_time[i]))
             else:
-                # 如果旁白中有换行符，即分为n段，则按照换行符进行分割，并间隔0.5s
-                text_split = text.split('\n')
-                # 如果旁白有对应的时间戳（是这段大旁白里的特定位置）
-                if start_time[i] is not None and end_time[i] is not None:
-                    cur_start = float(start_time[i])
-                    cur_end = float(end_time[i])
-                elif subtitle[i] is None:
-                    cur_start = float(end_time[i - 1]) + 0.1 if i > 0 else 0
-                    # 如果是最后一句旁白，后面没有字幕及时间戳了，就先把cur_end置为-1
-                    cur_end = float(
-                        start_time[i + 1]) if i + 1 < len(start_time) else -1
-                else:
-                    # 有字幕，可覆盖字幕
-                    cur_start = float(start_time[i])
-                    cur_end = float(end_time[i])
-                for x in text_split:
-                    if len(x) == 0:
-                        continue
-                    cur_end = max(cur_end, cur_start + (len(x) / normal_speed + normal_interval) / speed)
-                    narratage_text.append(x)
-                    narratage_start_time.append(cur_start)
-                    narratage_end_time.append(cur_end)
-                    cur_start = cur_start + (len(x) / normal_speed + normal_interval) / speed
-    return narratage_text, narratage_start_time, narratage_end_time
+                narratage_text.append(text)
+                """以下为之前自动根据表格生成旁白对应起始时间和终止时间的方法，目前不需要了
+                """
+                # # 如果旁白中有换行符，即分为n段，则按照换行符进行分割，并间隔0.5s
+                # text_split = text.split('\n')
+                # # 如果旁白有对应的时间戳（是这段大旁白里的特定位置）
+                # if start_time[i] is not None and end_time[i] is not None:
+                #     cur_start = float(start_time[i])
+                #     cur_end = float(end_time[i])
+                # elif subtitle[i] is None:
+                #     # 上一个字幕/旁白的终止时间后0.1s
+                #     cur_start = float(end_time[i - 1]) + normal_interval if i > 0 else 0
+                #     # 如果是最后一句旁白，后面没有字幕及时间戳了，就先把cur_end置为-1
+                #     cur_end = float(
+                #         start_time[i + 1]) if i + 1 < len(start_time) else -1
+                # else:
+                #     # 有字幕，可覆盖字幕
+                #     cur_start = float(start_time[i])
+                #     cur_end = float(end_time[i])
+                # for x in text_split:
+                #     if len(x) == 0:
+                #         continue
+                #     cur_end = max(cur_end, cur_start + (len(x) / (normal_speed * speed) + normal_interval))
+                #     narratage_text.append(x)
+                #     narratage_start_time.append(cur_start)
+                #     narratage_end_time.append(cur_end)
+                #     cur_start = cur_start + (len(x) / normal_speed + normal_interval) / speed
+            narratage_start_time.append(float(start_time[i]))
+            narratage_end_time.append(float(end_time[i]))
+            narratage_speed.append(speed)
+    return narratage_text, narratage_start_time, narratage_end_time, narratage_speed
 
 
 def second_to_str(seconds: float) -> str:
@@ -331,8 +338,7 @@ def ss_and_export(video_path: str, sheet_path: str, output_dir: str, speed: floa
     # print("read sheet at time: ", datetime.datetime.now())
     sheet_content = read_sheet(book_path)
     # print("get narratage text at time: ", datetime.datetime.now())
-    narratages, start_timestamp, end_timestamp = get_narratage_text(
-        sheet_content, speed)
+    narratages, start_timestamp, end_timestamp, cur_speed = get_narratage_text(sheet_content)
     # print("export caption at time: ", datetime.datetime.now())
     export_caption(sheet_content, caption_file)
     print("已导出字幕文件")
@@ -343,7 +349,7 @@ def ss_and_export(video_path: str, sheet_path: str, output_dir: str, speed: floa
     for i, text in enumerate(narratages):
         wav_path = root_path + '/%.2f.wav' % start_timestamp[i]
         narratage_paths.append(wav_path)
-        speech_synthesis(text, wav_path, chosen_speaker, speed)
+        speech_synthesis(text, wav_path, chosen_speaker, cur_speed[i])
         print("目前正在处理{}".format(wav_path))
         if state is not None:
             state[0] = float((i + 1) / len(narratages)) * 0.97