marouane53/transcribe ? reverse-engineered prompt

Build me a simple Python app that can transcribe audio and video files into text using Wit.ai.

I want it to handle YouTube links as well as local files like MP3, WAV, MP4, MKV, and AVI. If needed, it should automatically pull audio from a video, convert MP3 to WAV, then run the transcription and save the results as both TXT and SRT files. Make it feel easy to use from the terminal, with clear prompts for the user to choose a source and enter the file path or video link plus the language key.

Please also make it work nicely with a .env file for Wit.ai API keys, and include a Docker setup so it can be run without much local setup. If there are any current library or API details you need to check, look them up online.