Lecture Speech-to-Text Automation

I need a straightforward solution that listens to recorded or live lectures and turns everything that is said into clean, readable text. Accuracy matters more to me than real-time speed, but the final transcript should be well-punctuated and ready for immediate use in hand-outs or slides.

You are free to build this in the language or framework you prefer—Python with Whisper, Google Speech-to-Text, AWS Transcribe, or any engine you feel delivers the highest word-for-word precision for lecture-style audio. I will be providing sample recordings so we can fine-tune against typical classroom background noise and different speaker accents.

Deliverables
• A runnable script or small desktop/app package that accepts an audio or video file and outputs a full transcript
• Simple setup instructions plus a short README so I can install and run it on my own machine
• One test run on a sample lecture showing the achieved accuracy and any post-processing you perform (punctuation, paragraphing, timestamps if available)

If this sounds like your wheelhouse, tell me the engine you’d pick and why, plus a realistic turnaround for the first working draft.