diff --git a/main.py b/main.py index fd7d904..b9294a1 100644 --- a/main.py +++ b/main.py @@ -22,7 +22,7 @@ def main(): dictdir(file_dict_path, file_dict, "") global sound_dict sound_dict = get_sound_dict(srt_file_dict, file_dict, sound_type) - print(sound_dict) + # print(sound_dict) db = LiteDB() global_obj.set_value("sound_dict", sound_dict) global_obj.set_value("db", db) diff --git a/ui/guiclass.py b/ui/guiclass.py index cd6bab4..6084887 100644 --- a/ui/guiclass.py +++ b/ui/guiclass.py @@ -6,6 +6,7 @@ import pysrt from PySide6.QtCore import Qt from PySide6.QtGui import QFont +from pydub import AudioSegment from pydub.playback import play import global_obj @@ -211,10 +212,13 @@ def click_output_now(self): except: print("请输入以毫秒为单位的纯数字") else: - self.work_space_data.sound[start:end].export( + out_sound = self.work_space_data.sound[start:end] + out_sound = out_sound.set_frame_rate(16000).set_channels(1) # 设置为16k采样率，单声道 + out_sound.export( path + self.work_space_data.name + '_' + time.strftime("%Y_%m_%d_%H_%M_%S", time.localtime()) + ".wav", - format="wav") - print("导出当前文件为:" + path + self.work_space_data.name + '_' + time.strftime("%Y_%m_%d_%H_%M_%S", time.localtime()) + ".wav") + format="wav", bitrate="16k", codec='pcm_s16le') + print("导出当前文件为:" + path + self.work_space_data.name + '_' + time.strftime("%Y_%m_%d_%H_%M_%S", + time.localtime()) + ".wav") def click_back_to_main(self): """ 返回首页窗口 """ @@ -239,7 +243,40 @@ def to_inputfile(self): window3.show() def to_outputfile(self, name): + """ + 根据数据集名导出数据集，导出的路径为filepath/output/数据集名_日期时间/ + 音频路径为data/wav/数据集名_编号.wav + 文本路径为data/trans/sample.txt + """ print(f"导出数据集{name}") + path = "filepath/output" + path = path.replace("\\", "/") + if not path.endswith("/"): + path = path + "/" + path = path + name + '_' + time.strftime("%Y_%m_%d_%H_%M_%S", time.localtime()) + "/" + utils.check_mkdir(path + "data/wav/") + utils.check_mkdir(path + "data/trans/") + db = global_obj.get_value("db") + result_dict = db.select_output_data(name) + print(result_dict) + try: + all_sound = AudioSegment.from_file(result_dict["path"]).set_frame_rate(16000).set_channels(1) + except: + print(f"{result_dict['path']} 文件未找到，无法导出数据集") + + i = 0 + write_str = "" + for row in result_dict["data_list"]: + i = i + 1 + all_sound[row[1]:row[2]].export(path + "data/wav/" + f"{name}_" + str(i) + ".wav", format="wav", bitrate="16k", + codec='pcm_s16le') + write_str = write_str + row[0] + "\n" + + with open(path + "data/trans/sample.txt", "a", encoding="UTF-8") as f: + f.write(write_str) + is_delete = QMessageBox.information(self, "导出完成", f"导出数据集{name}成功\n请前往 {path} 查看", + QMessageBox.Yes,QMessageBox.Yes) + print(f"导出数据集{name}成功") def to_workspace(self, name): print(f"前往数据标注窗口 {name}") @@ -273,7 +310,7 @@ def table_refresh(self): window1.ui.tableWidget.setItem(row_count, 0, item1) window1.ui.tableWidget.setItem(row_count, 1, item2) window1.ui.tableWidget.setCellWidget(row_count, 2, self.button_workspace_for_row(str(row[0]), str(row[1]))) - window1.ui.tableWidget.setCellWidget(row_count, 3, self.button_output_for_row(str(row[0]))) + window1.ui.tableWidget.setCellWidget(row_count, 3, self.button_output_for_row(str(row[0]), str(row[1]))) window1.ui.tableWidget.setCellWidget(row_count, 4, self.button_delete_for_row(str(row[0]))) window1.ui.tableWidget.resizeColumnsToContents() # 表格列宽自动调整 @@ -292,9 +329,12 @@ def button_delete_for_row(self, name): input_btn.clicked.connect(lambda: self.is_delete_box(name)) return input_btn - def button_output_for_row(self, name): + def button_output_for_row(self, name, path): input_btn = QPushButton('导出数据集') input_btn.clicked.connect(lambda: self.to_outputfile(name)) + if not utils.is_sound_file_ok(path): + input_btn.setEnabled(False) + input_btn.setText("音频缺失") return input_btn def get_row_info(self): diff --git a/utils.py b/utils.py index 0178e69..e9fceef 100644 --- a/utils.py +++ b/utils.py @@ -64,21 +64,31 @@ def select_sound_path(self, name): for row in result: return row[0] - def select_dataset_data(self, name): + def select_dataset_data(self, name, num=4): """ 搜索某个数据集里的全部信息 """ c = self.conn.cursor() # result = c.execute(f"SELECT ROW_NUMBER() OVER(ORDER BY sound_start ASC)-1 AS xuhao ,sound_text,sound_start,sound_end,checked,can_use FROM {name} ORDER BY sound_start ASC") result = c.execute( - f"SELECT sound_id,sound_text,sound_start,sound_end,checked,can_use FROM {name} WHERE LENGTH(sound_text)>4 ORDER BY sound_start ASC") + f"SELECT sound_id,sound_text,sound_start,sound_end,checked,can_use FROM {name} WHERE LENGTH(sound_text)>{num} ORDER BY sound_start ASC") return list(result) - def select_dataset_row(self, name, id): + def select_dataset_row(self, name, id, num=4): """ 搜索数据集里的其中一条数据 """ c = self.conn.cursor() result = c.execute( f"SELECT sound_id,sound_text,sound_start,sound_end,checked,can_use FROM {name} WHERE sound_id = {id}") return list(result)[0] + def select_output_data(self, name, num=4): + """ 搜索指定数据集要导出的数据 """ + result_dict = {} + c = self.conn.cursor() + result = c.execute( + f"SELECT sound_text,sound_start,sound_end FROM {name} WHERE LENGTH(sound_text)>{num} AND checked=1 AND can_use=1 ORDER BY sound_start ASC") + result_dict["data_list"] = list(result) + result_dict["path"] = list(c.execute(f"SELECT sound_file_path FROM {name} limit 1"))[0][0] # 顺序很重要 + return result_dict + def insert_sound_line(self, input_list, path, name): """ 新增一条语音记录 """ c = self.conn.cursor() @@ -173,9 +183,9 @@ def listdir(path, list_name, file_end=""): # 传入存储的list def dictdir(path, dict_name, file_end=""): # 传入存储的dict """ 将目录下的文件名读存储在list中 - :param file_end: :param path: :param dict_name: + :param file_end: :return: """ for file in os.listdir(path):