GUI Showcase of using Whisper to transcribe and analyze Youtube video
This GUI is based on this article. But this article is quite old in 2024.
So i've changed a lot!
- PyQt5>=5.14
- openai
- python-docx
- psutil
- pytube - to install youtube video as an audio
- pydub
- requests
- openai-whisper - to extract the language and transcribe the content of the audio
- numpy<2.0.0
- ffmpeg (you can install this with choco install ffmpeg in Windows and sudo apt-get ffmpeg in linux)
- Actually the app includes ffmpeg.exe, so you don't need to download this.
- git clone ~
- pip install -r requirements.txt
- python main.py
First, This app will download the Youtube video as 128kb audio file.
Then this app trim the audio file with ffmpeg. The term "trim" means to remove the opening and ending music or silent portions from a video.
ffmpeg command will be run consequently after audio is downloaded.
Finally this app will transcribe the audio as verbose format, stream the output and display it in a text browser.
I use this video file as a sample. This is good sample video called "Microsoft (MSFT) Q4 2022 Earnings Call" which length is about 1 and a half hour
It only works in CUI for some reasons.