An R-based tool that uses Open-AIs whisper
to create transcripts from audio.
There are two ways how to use WhisperTranscribe
. Please see the requirements
first, before running the applications.
There is a shiny application in the WhisperTranscribe
sub-folder. The app is
located in the single app.R
file. By running this file the shiny app should
load.
The file transcribe.R
converts all file with the provided extension in the
input folder into transcripts in the output folder.
WhisperTranscribe needs Python + the Open-AI library. The setup file should install both of these. If you don't want to use the sandbox environment of reticulate you can install the required library using pip as follows:
pip install -U openai-whisper
See here https://github.com/openai/whisper for extended troubleshooting.
Whisper needs ffmpeg
to understand different audio formats.
# on Ubuntu or Debian
sudo apt update && sudo apt install ffmpeg
# on Arch Linux
sudo pacman -S ffmpeg
# on MacOS using Homebrew (https://brew.sh/)
brew install ffmpeg
# on Windows using Chocolatey (https://chocolatey.org/)
choco install ffmpeg
# on Windows using Scoop (https://scoop.sh/)
scoop install ffmpeg
If you are using conda and cannot find ffmpeg, it helps to directly install ffmpeg from the conda command line in the bin directory of the conda-environment by calling conda install ffmpeg
in the directory given by pyconfig()
from the test.R
.