Skip to content

Win10/11 Pre-packaged Version Download ​

macOS/Linux Installation Guide

Small donations help sustain open-source development

If this project helps you, please consider a small donation to support ongoing development and maintenance.

v4.18 Full Package (2.7G) ​

v4.18 Patch (400MB) ​

The patch cannot be used standalone — only for overlaying the existing full package. If the app crashes after patching, download the full package. If extracting the patch causes errors, download the full package.

Previous versions (v4.17 / v4.03)

If the new version has bugs, try the previous version.

v4.17 Full Package (2.7G)

v4.14 Full Package (2.7G)

v4.03 Full Package (2.7G)


Extraction notes:

  • Extract the 7z archive. After extraction, locate sp.exe and double-click to launch. Do NOT run inside the archive.
  • Do NOT extract to the desktop or system folders requiring admin permissions (e.g., Program Files on C drive).
  • Avoid path issues: extraction path must NOT contain Chinese characters, spaces, or special symbols.
  • If 7zip reports errors, try Bandizip

ChangeLog ​

  • feat: ask before start the TTS channel may not support the target language
  • fix(align): write the silent placeholder of a failed line at 48 kHz stereo
  • fix(dubbing): apply the volume setting in subtitle dubbing
  • fix(tts): convert Doubao 2.0 audio to 48 kHz stereo like other channels
  • fix(tts): write Doubao 2.0's mono PCM to both channels
  • fix(translator): stop a line ending in an ellipsis from joining the next
  • test(tts): expect the mono wav that the Doubao 2.0 channel now saves
  • fix(translator): strip multi-line <think> blocks before reading the translation
  • fix(align): in video slow-down, end the last slot where the video ends
  • fix(dubbing): import subtitles without [speaker] tags in multi-role dubbing again
  • fix(ui): take dropped files from their URLs and list their names
  • fix(prepare): escape the speaker name when removing its [Name] tag
  • fix(ui): convert an imported target-language .ass file to SRT
  • fix(params): stop adding deleted bundled reference audios back at every start
  • fix(srt): import and dub .ass files saved as GBK or UTF-16
  • fix(ui): save Multi-Speaker dubbing once, in the chosen folder
  • fix(dubbing): don't overwrite a same-named file with mp3/m4a output or one-request Edge-TTS
  • fix(dubbing): don't reuse cached clone dubbing from another video
  • fix(tts): save Qwen3-TTS lines re-dubbed with a preset voice
  • fix(tts): apply peak normalization and the silence settings when re-dubbing with built-in channels
  • fix(ui): use the channels' own params keys in the OpenRouter, DeepSeek and MiniMax windows
  • fix(recogn): list the whisper.cpp and Whisper.NET models again
  • fix(onlyone): start the dubbing proofreading panel from the main window's silence and sync choices
  • fix(ui): select the saved TTS model in the OpenRouter window
  • fix(tts): use the Index-TTS and VoxCPM versions chosen in the settings window again
  • fix(tts): apply the speech rate correctly in Piper and the GPT-SoVITS api.py request
  • fix(srt): import, dub and translate .srt and .txt files saved as GBK
  • fix(translator): reuse the translation cache when "Send SRT" is on
  • fix should_recogn2 when is_audio_trans
  • Optimize: allow add shibie.wav to target output as stt wav
  • Optimize dubbing peak normalize
  • STT proivder add audio.cpp