Practical overview
When it fits
Use it when you need Japanese subtitles from local media and want a workflow designed around noisy, long-form speech recognition rather than clean, short speech clips.
Who it is for
- People who want Japanese subtitles generated from locally held video or audio files.
- Users who need a graphical interface or command-line processing for individual files and folders.
- Developers exploring a transcription workflow that combines ASR with voice activity detection and post-processing.
Documented uses
- Generate Japanese subtitles from a video or audio file.
- Process a folder of media files and direct outputs to a selected directory.
- Save transcription output as SRT or WebVTT.
- Reduce recognition errors associated with non-speech vocalizations, silence and long recordings.
Requirements & limits
The repository is public and identifies the project under the MIT license. The README describes the application as free and locally run, with no cloud upload of media. Colab and Kaggle notebooks are also linked.
- The GUI and command-line interface both support transcription; the command line also documents whole-folder processing and output selection.
- SRT is the default output; WebVTT and both formats are supported.
- Windows has a standalone installer. Source-install instructions are also provided for Windows, Apple Silicon macOS and Linux.
- The project recommends 8 GB RAM and 8 GB disk as minimums, and 16 GB RAM and 15 GB disk with models as recommended.
- The Windows installer downloads dependencies, and the first transcription downloads about 3 GB of models.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Generate subtitles from a file using the command line
Transcribe a supported audio or video file and save subtitles beside it or in a chosen output directory.
Before you start
- WhisperJAV installed and its speech-recognition models available.
- An input format supported by FFmpeg.
Steps
Run
whisperjav video.mp4to use the defaults, or provide a file path and documented mode or sensitivity options.For a folder, run
whisperjav /path/to/folder --output-dir ./subtitles.Use the default SRT output, or specify
--output-format bothfor SRT and WebVTT.
Expected outcome
Subtitle files are produced in SRT by default, or in both SRT and WebVTT when requested.
Guide limitations
- Transcription quality varies with source audio.
- CPU processing is supported but can be substantially slower than GPU processing.
Install WhisperJAV on Windows with the standalone installer
Prepare the Windows application and launch its graphical interface.
Before you start
- Windows 10 or later.
- The standalone installer downloaded from the project Releases page.
- Enough time for setup and model downloads; first transcription downloads approximately 3 GB of models.
Steps
Download the
.exefrom the Releases page.Run the installer; the project says administrator rights are not required.
Wait while it sets up Python, PyTorch, FFmpeg and dependencies and detects the available NVIDIA driver or CPU-only configuration.
Launch WhisperJAV from the desktop shortcut.
Expected outcome
The desktop application is installed and ready to begin a transcription after models are downloaded.
Guide limitations
- Installation may take 10–20 minutes.
- CPU-only processing is slower; the project recommends an NVIDIA RTX 2060 or newer for GPU use.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 11 October 2026
- Original website title
- meizhong986/WhisperJAV · Official homepage title · retained in its source language