Whisper transcriptions

The main idea here is to create a tool to transcript audios to text by using Whisper (speech to text from OpenAI)

For a quick test enter to https://whisper-transcriptions.streamlit.app/ and give it a try

If you want to be able to change it or run it locally, then read the following steps

It doesn't matter how you are going to transcript, first you should do this

1. You should install python3.8 :)

2. on Ubuntu or Debian:

    sudo apt update && sudo apt install ffmpeg
   

Then to run it

1. Create virtualenv using python3 (follow https://virtualenvwrapper.readthedocs.io/en/latest/install.html)

    virtualenv venv -p python3.8
   

2. Activate the virtualenv

    source env/bin/activate
   

3. Install python requirements (be patient, it takes time)

    pip install -r requirements.txt
   

If you want to run script for bulk transcription

Run

    python transcript_automation.py PATH_TO_FOLDER_WITH_MP3s

For every mp3 file you'll get a text file with audio transcription in the same folder where audio is. Text file will be named as the audio file.

If you want to run it with Streamlit (Work In Progress)

4. Run main.py with streamlit and get a transcription!

    streamlit run main.py