here's a concise guide on installing and using "whisper" locally for automatic subtitle generation from videos.

README.md

0. Requirements

• video file
• conda package manager(installed)
• you

1. How to?

about whisper

check this github repo.

1. make virtual env

conda create -n whisper python=3.9

2. activate virtual env

conda activate whisper

3. install pytorch If you want to do it faster. you need to install pytorch-gpu

conda install pytorch-gpu -y

4. check cuda availablity

• open the python interpreter.

python

• in python interpreter.

import torch
torch.cuda.is_available()
exit()

it must print true

5. then, install whisper

pip install git+https://github.com/openai/whisper.git

6. in terminal

whisper [Wang.mp4] --model large

• [Wang.mp4] is your path for video.
• It take 7min 11sec for 40min video(I used RTX4090)
• pretty cool

2. Result

when it is done. deactivate virtual env.

conda deactivate

In same directory of video, there is .srt, .txt files in it.