fix: bugs and update

README.md

@@ -240,6 +240,23 @@ apt-get install ffmpeg
 
 
 
+#  源码部署问题说明
+
+1. 默认使用 ctranslate2的4.x版本，仅支持CUDA12.x版本，如果你的cuda低于12，并且无法升级cuda到12.x，请执行命令卸载ctranslate2然后重新安装
+
+```
+
+pip uninstall -y ctranslate2
+
+pip install ctranslate2==3.24.0
+
+```
+
+2. 可能会遇到 `xx module not found ` 之类错误，请打开 requirements.txt，搜索该 xx 模块，然后将xx后的 ==及等会后的版本号去掉
+
+
+
+
 # 使用教程和文档
 
 请查看 https://pyvideotrans.com/guide.html

ffmpeg下载.txt

@@ -0,0 +1,5 @@
+windows������ffmpeg�󣬽�binĿ¼�µ� ffmpeg.exe��ffprobe.exe �ļ����Ƶ���ǰĿ¼�µ� ffmpeg �ļ�����
+
+���ص�ַ
+
+https://github.com/BtbN/FFmpeg-Builds/releases/download/autobuild-2023-11-30-12-55/ffmpeg-n6.0.1-win64-gpl-6.0.zip

requirements.txt

@@ -20,7 +20,7 @@ click==7.1.2
 colorama==0.4.6
 coloredlogs==15.0.1
 contourpy==1.1.1
-ctranslate2==3.24.0
+ctranslate2
 cycler==0.12.1
 decorator==4.4.2
 deepl==1.16.1

videotrans/__init__.py

@@ -1,4 +1,4 @@
 # -*- coding: utf-8 -*-
 
-VERSION="v1.78"
-VERSION_NUM=110708
+VERSION="v1.79"
+VERSION_NUM=110709

videotrans/box/win.py

@@ -21,6 +21,7 @@
 import shutil
 from videotrans.ui.toolboxen import Ui_MainWindow
 from videotrans.util.tools import get_azure_rolelist, get_edge_rolelist
+from pathlib import Path
 
 
 class MainWindow(QMainWindow, Ui_MainWindow):
@@ -120,6 +121,7 @@ def initUI(self):
         self.hecheng_role.addItems(['No'])
         self.hecheng_language.currentTextChanged.connect(self.hecheng_language_fun)
         self.hecheng_startbtn.clicked.connect(self.hecheng_start_fun)
+        self.listen_btn.clicked.connect(self.listen_voice_fun)
         self.hecheng_opendir.clicked.connect(lambda: self.opendir_fn(self.hecheng_out.text().strip()))
         # 设置 tts_type
         self.tts_type.addItems([i for i in config.params['tts_type_list']])
@@ -238,11 +240,24 @@ def fanyi_import_fun(self, obj=None):
 
     def hecheng_import_fun(self):
         fnames, _ = QFileDialog.getOpenFileNames(self, "Select srt", config.last_opendir,
-                                                 "Text files(*.srt)")
+                                                 "Text files(*.srt *.txt)")
         if len(fnames) < 1:
             return
         for (i, it) in enumerate(fnames):
-            fnames[i] = it.replace('\\', '/').replace('file:///', '')
+            it=it.replace('\\', '/').replace('file:///', '')
+            if it.endswith('.txt'):
+                shutil.copy2(it,f'{it}.srt')
+                # 使用 "r+" 模式打开文件：读取和写入
+                with open(f'{it}.srt', 'r+', encoding='utf-8') as file:
+                    # 读取原始文件内容
+                    original_content = file.readlines()
+                    # 将文件指针移动到文件开始位置
+                    file.seek(0)
+                    # 将新行内容与原始内容合并，并写入文件
+                    file.writelines(["1\n","00:00:00,000 --> 05:00:00,000\n"] + original_content)
+
+                it+='.srt'
+            fnames[i] = it
 
         if len(fnames) > 0:
             config.last_opendir = os.path.dirname(fnames[0])
@@ -649,6 +664,73 @@ def disabled_shibie(self, type):
         self.shibie_language.setDisabled(type)
         self.shibie_model.setDisabled(type)
 
+
+    # 试听配音
+    def listen_voice_fun(self):
+        lang = translator.get_code(show_text=self.hecheng_language.currentText())
+        if not lang or lang=='-':
+            return QMessageBox.critical(self, config.transobj['anerror'], "选择字幕语言" if config.defaulelang=='zh' else 'Please target language')
+        text = config.params[f'listen_text_{lang}']
+        role = self.hecheng_role.currentText()
+        if not role or role == 'No':
+            return QMessageBox.critical(self, config.transobj['anerror'], config.transobj['mustberole'])
+        voice_dir = os.environ.get('APPDATA') or os.environ.get('appdata')
+        if not voice_dir or not Path(voice_dir).exists():
+            voice_dir = config.rootdir + "/tmp/voice_tmp"
+        else:
+            voice_dir = voice_dir.replace('\\', '/') + "/pyvideotrans"
+        if not Path(voice_dir).exists():
+            Path(voice_dir).mkdir(parents=True,exist_ok=True)
+        lujing_role = role.replace('/', '-')
+
+        rate = int(self.hecheng_rate.value())
+        tts_type = self.tts_type.currentText()
+
+        if rate >= 0:
+            rate = f"+{rate}%"
+        else:
+            rate = f"{rate}%"
+        volume=int(self.volume_rate.value())
+        pitch=int(self.pitch_rate.value())
+        volume=f'+{volume}%' if volume>=0 else f'{volume}%'
+        pitch=f'+{pitch}Hz' if pitch>=0 else f'{volume}Hz'
+
+
+
+        voice_file = f"{voice_dir}/{config.params['tts_type']}-{lang}-{lujing_role}-{volume}-{pitch}.mp3"
+        if config.params['tts_type'] == 'GPT-SoVITS':
+            voice_file += '.wav'
+
+        obj = {
+            "text": text,
+            "rate": '+0%',
+            "role": role,
+            "voice_file": voice_file,
+            "tts_type": config.params['tts_type'],
+            "language": lang,
+            "volume":volume,
+            "pitch":pitch,
+        }
+
+        if config.params['tts_type'] == 'clone-voice' and role == 'clone':
+            return
+        # 测试能否连接clone
+        if config.params['tts_type'] == 'clone-voice':
+            try:
+                tools.get_clone_role(set_p=True)
+            except:
+                QMessageBox.critical(self, config.transobj['anerror'],
+                                     config.transobj['You must deploy and start the clone-voice service'])
+                return
+
+        def feed(d):
+            QMessageBox.critical(self, config.transobj['anerror'], d)
+
+        from videotrans.task.play_audio import PlayMp3
+        t = PlayMp3(obj, self)
+        t.mp3_ui.connect(feed)
+        t.start()
+
     # tab-4 语音合成
     def hecheng_start_fun(self):
         txt = self.hecheng_plaintext.toPlainText().strip()
@@ -700,6 +782,10 @@ def hecheng_start_fun(self):
 
 
         issrt = self.tts_issrt.isChecked()
+        if len(self.hecheng_files)>0:
+            self.tts_issrt.setChecked(True)
+            issrt=True
+
         self.hecheng_task = WorkerTTS(self,
                                       files=self.hecheng_files if len(self.hecheng_files) > 0 else txt,
                                       role=role,

videotrans/box/worker.py

@@ -480,9 +480,9 @@ def merge_audio_segments(self, *, segments=None, queue_tts=None, video_time=0, o
             if diff >= 0:
                 it['end_time'] += diff
                 offset += diff
-            else:
+            #else:
                 # 配音小于原时长，添加静音
-                merged_audio += AudioSegment.silent(duration=abs(diff))
+            #    merged_audio += AudioSegment.silent(duration=abs(diff))
 
             if i > 0:
                 silence_duration = it['start_time'] - queue_tts[i - 1]['end_time']
@@ -495,10 +495,10 @@ def merge_audio_segments(self, *, segments=None, queue_tts=None, video_time=0, o
             queue_tts[i] = it
             merged_audio += segment
 
-        if video_time > 0 and (len(merged_audio) < video_time):
+        #if video_time > 0 and (len(merged_audio) < video_time):
             # 末尾补静音
-            silence = AudioSegment.silent(duration=video_time - len(merged_audio))
-            merged_audio += silence
+        #    silence = AudioSegment.silent(duration=video_time - len(merged_audio))
+        #    merged_audio += silence
         # 创建配音后的文件
         try:
             merged_audio.export(out, format="wav")

videotrans/recognition/all.py

@@ -72,6 +72,8 @@ def recogn(*,
             if config.exit_soft or (config.current_status != 'ing' and config.box_recogn != 'ing'):
                 del model
                 return None
+            if not segment.words or len(segment.words)<1:
+                continue
             sidx += 1
             start = int(segment.words[0].start * 1000)
             end = int(segment.words[-1].end * 1000)

videotrans/ui/toolboxen.py

@@ -247,6 +247,13 @@ def setupUi(self, MainWindow):
         self.hecheng_role.setObjectName("hecheng_role")
         self.formLayout_4.setWidget(0, QtWidgets.QFormLayout.FieldRole, self.hecheng_role)
         self.horizontalLayout_10.addLayout(self.formLayout_4)
+
+        self.listen_btn = QtWidgets.QPushButton(self.tab_2)
+        self.listen_btn.setFixedWidth(80)
+        self.listen_btn.setToolTip(config.uilanglist.get("shuoming01"))
+        self.listen_btn.setText(config.uilanglist.get("Trial dubbing"))
+        self.horizontalLayout_10.addWidget(self.listen_btn)
+
 
 
         self.formLayout_5 = QtWidgets.QFormLayout()

videotrans/util/tools.py

@@ -714,12 +714,22 @@ def get_subtitle_from_srt(srtfile, *, is_file=True):
                 raise Exception(f'get srtfile error:{str(e)}')
     else:
         content = srtfile.strip().splitlines()
+    # remove whitespace
+    content=[c for c in content if c.strip()]
+
     if len(content) < 1:
-        raise Exception("srt content is 0")
-
+        raise Exception("srt content is empty")
+    
     result = format_srt(content)
+
+    # txt 文件转为一条字幕
     if len(result) < 1:
-        return []
+        if is_file and srtfile.endswith('.txt'):
+            result= [
+                {"line":1,"time":"00:00:00,000 --> 05:00:00,000","text":"\n".join(content)}
+            ]
+        else:
+            return []
 
     new_result = []
     line = 1