Transcripter: Video to SRT using FFmpeg & OpenAI Whisper

Transcripter is a tool that generates .srt subtitle files from video files using OpenAI's Whisper for transcription. It also supports translation via Argos Translate to convert subtitles into different languages.

The app provides several settings to fine-tune the quality and accuracy of transcriptions.

---

Features

Automatic Speech-to-Text: Converts video audio into text with OpenAI Whisper. SRT Subtitle Generation: Saves transcriptions in .srt format with proper timestamps. Multilingual Support: Uses Argos Translate to translate subtitles into different languages. Customizable Settings: Adjust model parameters for optimal transcription quality. Automatic Dependency Installation: Downloads and sets up required software in a virtual environment.

---

Requirements

• Python: minimum 3.9.5
• Storage: Up to 5GB free space (if using the large Whisper model).
• GPU Acceleration (Optional):
  • CUDA 12.2 required for GPU processing.
  • Falls back to CPU if no compatible GPU is detected.
• Operating System:
  • Windows (Supported)
  • macOS (Planned Apple Silicon Support)

---

Installation & Usage

1. Copy the entire project folder to any location on your system.
2. Run create_desktop_shortcut.py to create a desktop shortcut, then launch the app from the created icon.

• ⚠️ The first launch may take a few moments while it downloads and installs all dependencies. Alternatively, you can run launcher.py manually with Python:

  python launcher.py

3. Select a movie file, select the target language and Whisper model.
4. If Force new srt is checked, it will process the video and transcript it even if a subtitle already exists. If unchecked and a subtitle exists, it will go to translate mode only (faster).

Notes:

• Settings are stored in /users/USERNAME/.transcripter_prefs You can tweak the settings and click on "save as default" to keep those settings for the next session or click on "reset all" to reset them default button.
• Most of the needed package will be installed into /users/USERNAME/documents/transcripter_venv
• Whisper models will be downloaded on request into /users/USERNAME/.cache
• ArgosTranslate models will be downloaded on request into /users/USERNAME/.local

Todo:

• Improve the progress bar feedback
• Add support for macOS