https://github.com/valkryst/whisper_automations

Various scripts for automating tasks using OpenAI's Whisper.
https://github.com/valkryst/whisper_automations

automation openai subtitle subtitle-generator transcription translation whisper

Last synced: 8 days ago
JSON representation

Various scripts for automating tasks using OpenAI's Whisper.

• Host: GitHub
• URL: https://github.com/valkryst/whisper_automations
• Owner: Valkryst
• License: mit
• Created: 2023-03-09T02:56:11.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2023-11-29T01:40:49.000Z (12 months ago)
• Last Synced: 2023-11-29T02:30:43.065Z (12 months ago)
• Topics: automation, openai, subtitle, subtitle-generator, transcription, translation, whisper
• Language: Shell
• Homepage: https://www.valkryst.com/posts?search=Whisper
• Size: 14.6 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
## Installation

All of these scripts assume that you have followed [my guide](https://www.valkryst.com/posts/42) to installing and using Whisper. These scripts should be placed alongside the `whisper`, `whisper-large-v3`, and `whisper.cpp` repository folders.

To download these scripts and their dependencies, copy and paste the following commands into your bash terminal.

```shell

# Install Dependencies

sudo apt update

sudo apt install ffmpeg wget

# Download Scripts

wget https://github.com/Valkryst/Whisper_Automations/blob/main/extract_audio.sh

wget https://github.com/Valkryst/Whisper_Automations/blob/main/generate_subtitles.sh

wget https://github.com/Valkryst/Whisper_Automations/blob/main/regenerate_model.sh

# Modify the script permissions to allow them to be run.

chmod +x extract_audio.sh

chmod +x generate_subtitles.sh

chmod +x regenerate_model.sh

```

## Usage

1. If it has been some time since you have pulled the latest `whisper-large-v3` model, you can run `./regenerate_model.sh` to update and regenerate it.

2. Copy your video files into the same directory as the scripts, and run `./extract_audio.sh`. This will extract _only_ the first audio track of each video file in the directory, and it will convert that audio track to a 16kHz _.wav_ file as required by Whisper.

3. Run `./generate_subtitles.sh` to generate an `.srt` file for each of the `.wav` files. This may take a significant amount of time.

## Notes

* The `generate_subtitles.sh` script will attempt to automatically detect the source file's language and, if necessary, translate it to English.

* The subtitles will never be perfect, but they do seem to work reasonably well.