1. 修改判定字幕的算法; 2.在界面中添加停止按钮，用于终止当前任务，重新开始新的任务; 3.将检测旁白的进度条优化为真进度条;

detect_with_asr.py

@@ -73,7 +73,7 @@ def concat_wav(root):
     return output_file
 
 
-def detect_with_asr(video_path, book_path, start_time=0, end_time=-1):
+def detect_with_asr(video_path, book_path, start_time=0, end_time=-1, state=None):
     # 临时存储各种中间产物的文件夹
     tmp_root = './tmp'
     if not os.path.exists(tmp_root):
@@ -102,9 +102,9 @@ def detect_with_asr(video_path, book_path, start_time=0, end_time=-1):
 
     sys.path.append("./PaddlePaddle_DeepSpeech2")
     from infer_path import predict_long_audio_with_paddle
-    table_content = predict_long_audio_with_paddle(audio_path, book_name_xlsx, start_time)
+    table_content = predict_long_audio_with_paddle(audio_path, start_time, state)
     write_to_sheet(book_name_xlsx, sheet_name_xlsx, table_content)
-
+    state[0] = 1
     # 删除中间文件
     # shutil.rmtree(tmp_root)
 

detect_with_ocr.py

@@ -14,7 +14,7 @@ up_b, down_b = 0, 0
 ocr = PaddleOCR(use_angle_cls=True, lang="ch", show_log=False)
 
 
-def get_position(video_path,start_time):
+def get_position(video_path, start_time):
     """
     根据对视频中的画面进行分析，确定字幕的位置，以便后续的字幕识别
     :param video_path: 视频存储路径
@@ -130,7 +130,7 @@ def detect_subtitle(img):
     return None
 
 
-def process_video(video_path, begin, end):
+def process_video(video_path, begin, end, state):
     """
     处理视频，主要完成对字幕的捕捉以及根据字幕分析得出旁白可能位置的任务
     :param video_path: 待处理视频的路径
@@ -156,6 +156,7 @@ def process_video(video_path, begin, end):
         cnt += 1
         # 每秒取4帧画面左右
         if cnt % int(fps / 4) == 0:
+            state[0] = float(cnt / video.get(cv2.CAP_PROP_FRAME_COUNT)) if state[0] < 0.99 else 0.99
             subTitle = detect_subtitle(frame)
             # 第一次找到字幕
             if lastSubTitle is None and subTitle is not None:
@@ -215,7 +216,7 @@ def write_excel_xlsx(path, sheet_name, value):
     workbook.save(path)
 
 
-def detect_with_ocr(video_path, book_path, start_time, end_time):
+def detect_with_ocr(video_path, book_path, start_time, end_time, state):
     book_name_xlsx = book_path
     sheet_name_xlsx = "旁白插入位置建议"
 
@@ -225,10 +226,11 @@ def detect_with_ocr(video_path, book_path, start_time, end_time):
 
     # 获取并构建输出信息
     table_head = [["起始时间", "终止时间", "字幕", '建议', '解说脚本']]
-    table_content = table_head + process_video(video_path, start_time, end_time)
+    table_content = table_head + process_video(video_path, start_time, end_time, state)
 
     # 输出旁白位置推荐信息到表格
     write_excel_xlsx(book_name_xlsx, sheet_name_xlsx, table_content)
+    state[0] = 1
 
 
 if __name__ == '__main__':

judge_subtitle.py

@@ -68,7 +68,6 @@ def detect_movie(video_path, start, interval):
     interval = interval * fps
     random_number = 50
     ans = [False] * 3
-    print(ans)
     for i in range(3):
         random_list = random_int_list(start, start + interval, random_number)
         start = start + interval
@@ -82,15 +81,20 @@ def detect_movie(video_path, start, interval):
             if ans[i]:
                 print(random_point)
                 break
+        if not ans[i]:
+            print('{}-{}时间段内未检测到字幕'.format(start, start + interval))
+        if i == 1 and Counter(ans).most_common(1)[0][0] is False:
+            break
     video.release()
     print(ans)
     return Counter(ans).most_common(1)[0][0]
 
 
 if __name__ == '__main__':
-    video_path = r'D:\heelo\hysxm.mp4'
-    start_time = time.time()
-    start = 90
-    interval = 120
-    print(detect_movie(video_path, start, interval))
-    print(time.time() - start_time)
+    pass
+    # video_path = r'D:\heelo\hysxm.mp4'
+    # start_time = time.time()
+    # start = 90
+    # interval = 120
+    # print(detect_movie(video_path, start, interval))
+    # print(time.time() - start_time)

narratage_detection.py

@@ -18,7 +18,7 @@ def trans_to_seconds(timepoint):
     return time_in_seconds
 
 
-def detect(video_path, start_time, end_time, book_path):
+def detect(video_path, start_time, end_time, book_path, state):
     print("开始检测")
     if book_path is None:
         book_path = os.path.basename(video_path).split('.')[0] + ".xlsx"
@@ -30,9 +30,9 @@ def detect(video_path, start_time, end_time, book_path):
 
     has_subtitle = detect_movie(video_path, start_time, 60)
     if has_subtitle:
-        detect_with_ocr(video_path, book_path, start_time, end_time)
+        detect_with_ocr(video_path, book_path, start_time, end_time, state)
     else:
-        detect_with_asr(video_path, book_path, start_time, end_time)
+        detect_with_asr(video_path, book_path, start_time, end_time, state)
 
 
 if __name__ == '__main__':

speech_synthesis.py

@@ -160,6 +160,7 @@ def ss_and_export(sheet_path, output_dir, speed, caption_file, state):
     for i, text in enumerate(narratages):
         wav_path = os.path.join(root_path, '%.2f.wav' % start_timepoint[i])
         speech_synthesis(text, wav_path, speed)
+        time.sleep(1)
         print("目前正在处理{}".format(wav_path))
         state[0] = float((i + 1) / len(narratages))