An AI captioning tool written with inversify.js. Capsaicin is capable of either generating whisper captions for a given audio file or converting an .srt captions file into a series of images.
This script requires that you install the OpenAI Whisper cli, and ImageMagick. If you do not wish to use Whisper functionality or vice versa for the image outputs this step can be neglected. You will also need to have NodeJS and yarn installed. Then see scripts for instructions on how to build.
capsaicin [options] [filenames...]
| Flag | Purpose | Default | Choices |
|---|---|---|---|
-V, --version |
output the version number | ||
--outdir <path> |
The directory to output files to | ./capsaicin | |
-w, --whisper |
Generate captions for [filenames…] with OpenAI Whisper | ||
-c, --convert |
Convert given caption file(s) to the supplied format | ||
-p, --png |
Export captions as PNG images | ||
-j, --jpg |
Export captions as JPEG images | ||
-g, --gif |
Export captions as GIF images | ||
-t, --tiff |
Export captions as TIFF images | ||
--noCUDA |
Disable CUDA support (try this option if you get Whisper errors) | ||
--im-size <size> |
The -size input provided to ImageMagick | 1920x1080 | |
--im-pointsize <pointsize> |
The -pointsize input provided to ImageMagick | 48 | |
--im-gravity <gravity> |
The -gravity input provided to ImageMagick | South | magick -list gravity |
--im-font <font> |
The -font input provided to ImageMagick | Cubano-Normal | magick -list font |
--model <model> |
The model for Whisper to use | whisper –help | |
-h, --help |
Display helptext |
Capsaicin provides the following scripts, which can be launched through yarn name_of_script:
run— run test suitesui— run test server-
dev— run development server, which will build the typescript upon save -
build— compile the typescript to javascript -
start— launch the compiled javascript