essanousy/azure_speech_ui ? reverse-engineered prompt

Build me a simple Streamlit app that lets me upload a WAV file and send it to Azure Speech batch transcription with speaker separation. I want it to use credentials from a local .env file, store the audio in Azure Blob Storage, submit the transcription job, and keep checking the status until it finishes. Then it should download the results into a local outputs folder as both a readable text file and a JSON file with the transcript details, speaker labels, timing, and an estimated cost based on the audio length. Please include a clean job list so I can see previous uploads, refresh statuses manually or automatically, and avoid creating duplicate jobs for the same file. Use French as the default language and support up to 7 speakers. If you need to, look up the current Azure docs online so the API flow is correct.