OpenAI recently released "Whisper" - a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multi-task model that can perform multilingual speech recognition as well as speech translation and language identification.
I have been experimenting with it and found it's accuracy to be amazing.
Here is the python file I have been using to transcribe my podcasts to readable format.
Full OpenAI whisper installation instructions (PIP, ffmpeg) here: