fix: Enhance transcription robustness and UI display in app.js

Description: This update is a comprehensive cleanup of the ASR (Automatic Speech Recognition) frontend. It resolves technical problems ranging from "freezing" performance issues.

What we did: Added logic to automatically detect the correct file extension (like .ogg or .m4a) from the browser's audio recording before sending it to the server.

How it fixed the issue: Previously, sending a generic filename would sometimes cause the ASR model to fail because it couldn't identify how to decode the audio. Now, the backend always receives the correct file type, ensuring compatible transcription.

Closes #2 (closed)

Before:

After: