Whisper return Wrong text

Question

I am using whisper to transcribe audio data , i have run local server by flask. but after first request that save and return me old text too .`

    def transcribe(self, audio):
        result = self.speech2text_model.transcribe(audio)
        results = result['segments']
        result = [{'text': item['text'], 'start': item['start'], 'end': item['end']} for item in results]

        return result

 

 

how can i solve that problem?  I want to  get new response every time 

i have tried `--condition_on_previous_text =False"` it doesn't work 
also `--compression_ratio_threshold` couldn't change that