Skip to content

0.5.0

Compare
Choose a tag to compare
@CheshireCC CheshireCC released this 18 Nov 08:12
· 41 commits to main since this release

0.5.0

0.5.0 改动

  • 重构模型参数页面 UI 布局设计
    • 弃用转换模型功能
    • 重新设计参数项布局
  • 其他 UI 优化
  • 将输出文件编码参数应用于更多输出文件
    • 现在所有输出文件格式: .srt.vtt.txt.lrc.smi 格式的输出文件都可以设置文件编码而不仅仅是 .srt 格式。
  • 添加 设置 界面
    • 添加保存软件各项配置的设置项
    • 添加设置自动加载模型的设置项 #33
    • 添加设置页面清除软件临时存储的功能按钮
    • 添加设置页面打开临时存储目录的功能按钮
    • 添加设置页面选择转写完成之后是否自动跳转页面的设置项 #38
    • 添加自动清除临时文件设置项
    • 添加打开 log 文件的按钮
    • 添加语言设置选项 #34
  • 添加自动保存软件配置的功能,在 设置 页面设置是否自动保存
    • 添加自动保存、加载主题设置的功能 #38
    • 添加自动保存、加载模型参数配置功能
    • 添加自动保存、加载 VAD 参数配置的功能
    • 添加自动保存、加载 转写参数配置的功能
    • 添加自动保存、加载 Demucs 参数配置的功能
    • 添加自动保存、加载 字幕表格样式参数配置的功能
    • 添加自动保存、加载 whisperX 参数配置的功能
  • 添加自动加载模型的功能
    • 如果 设置 页面下的 自动加载模型 选项被开启,软件启动后将会按照 模型参数 配置自动加载模型
    • 该功能需要正确保存前次 模型参数 配置的前提下开启,所以要使用该功能,自动保存配置 选项必须开启。
  • 添加更多 log 信息
    - 添加 faster_whisper 的详细日志 faster_whisper.log 文件
  • 取消文件转写时,如果已经有转写结果,将输出结果显示到输出页面
  • 将输入文件检测设置成多线程运行避免文件过多时界面假死以及信息窗口不能弹出
    - 启用子线程进行输入文件内容检测
  • 修正加载页面拼写错误 #38
  • 最终,我为 huggingface 用户令牌参数 找到了家
  • 调整输出文件的逻辑,
    - 将会使用全局变量保存当前活动的转写结果,所有功能均可单独工作并输出结果,包括 whisperX 的时间戳对齐和说话人分离功能
  • 修复连续转写操作下会闪退的 bug
  • 修复转写结束之后可能出现的闪退 bug

提示

  • 如果 whisperX 功能出现异常且 log 文件显示 Error: [WinError 2] 系统找不到指定的文件。 请确保已经正确安装 ffmpeg 。如果没有安装过 ffmpeg 可以在本次发布中下载 ffmpeg.zip 解压之后将整个文件夹放入软件安装目录
  • 百度网盘更新地址:https://pan.baidu.com/s/18Yq6pH_6KB_Ht4U03AgkZA?pwd=hbie 欢迎订阅

0.5.0 Changes

  • UI layout Design of reconstructed Model Parameter Page

    • Deprecate the transformation model function

    • Redesign the layout of parameter items

  • Other UI optimizations

  • Apply output file encoding parameters to more output files

    • Now all output files in the formats of .srt, .vtt, .txt, .lrc and .smi can be encoded instead of just .srt.
  • Add Settings interface

    • Add settings that save the configuration of the software

    • Add setting item # 33 that sets the automatic loading of the model

    • Add settings page to clear the function button of the temporary storage of the software

    • Add settings page to open the function button for temporary storage directory

    • Add the settings page to select whether to automatically jump to the page after the conversion is completed. # 38

    • Add automatic cleanup of temporary file settings

    • Add a button to open the log file

    • Add language setting options # 34

  • Add the function to automatically save the software configuration, and whether the settings on the Settings page are automatically saved.

    • Add the function of automatically saving and loading theme settings # 38

    • Add auto-save and load model parameter configuration function

    • Add the function of automatically saving and loading VAD parameter configuration

    • Add the function of automatically saving and loading transfer parameter configuration

    • Add the function of automatically saving and loading Demucs parameter configuration

    • Add the function of automatically saving and loading subtitle table style parameter configuration

    • Add the function of automatically saving and loading whisperX parameter configuration

    • Add the ability to load models automatically

      • If the auto load model option under the Settings page is enabled, the software will automatically load the model according to the model parameter configuration after startup.

      • This feature needs to be enabled on the premise that the previous model parameter configuration is saved correctly, so the auto-save configuration option must be enabled to use this feature.

  • Add more log information

    • Add a detailed log faster_whisper.log file for faster_whisper
  • When canceling file transfer, if there is already a transfer result, the output result is displayed to the output page

  • Set the input file detection to multithreaded to avoid the interface dying when there are too many files and the information window cannot pop up.

    • Enable child threads to detect the contents of input files
  • Fixed the misspelling of the loaded page # 38

  • Finally, I found a home for the parameter huggingface user token

  • Adjust the logic of the output file

    • Global variables will be used to save the transcription results of the current activity, and all functions can work independently and output the results, including whisperX timestamp alignment and speaker separation
  • Fixed bug that flickered during continuous write operations

  • Fixed possible flashback bug after the end of the overwrite

tips

  • If the whisperX function is abnormal and the log file shows Error: [WinError 2] the system cannot find the specified file. Please make sure that ffmpeg is installed correctly. If you have not installed ffmpeg, you can download ffmpeg.zip in this release and put the entire folder into the software installation directory after decompression.