Current Prompt:
Make the conversation transcription error free and use the conversation to infer a short phrase as a query to search after the conversation. And tell whether the speakers are looking for new information beyond the conversation that they're having.
Answer in JSON format, with these keys - 'corrected_conversation','key_phrase' and 'seeking_new_info'.