Decoding Wisper API response for transcribe

API
1

Hello, I am testing the whisper API to transcribe Marathi ( an language spoken in India & is supported by by API) given beow is the code

import openai
openai.api_key = "my key"
audio_file= open("C:\\Users\\pc\\Downloads\\sample.mp4", "rb")
transcript = openai.Audio.transcribe("whisper-1", audio_file,"response_format : utf-8","language:mr")
print(transcript)
print(transcript["text"].encode('utf-16').decode('utf-16'))

the problem is as follows

  1. The response is a unicode in text format
    {
    “text”: “\u0906\u091c \u0906\u092a\u0928\u093e \u0906\u0930\u093e\u092e \u0915\u0930\u094b”
    }
    hence i used encode-decode,but is this method correct
    आज आपना आराम करो
  2. Using the same audio i did translate to english i got an accurate translation
    {
    “text”: “It’s a holiday today. Let’s rest.”
    }
    but the transcribe skips " Its a holiday today" and transcribes only "Lets Rest "