app.py

import gradio as gr
from huggingface_hub import InferenceClient
import os

listen_client = InferenceClient("openai/whisper-medium")

def func(audio_):
    val=listen_client.automatic_speech_recognition(audio_).text
    return val
    
gr.Interface(fn=func, inputs=gr.Audio(sources=["microphone"], type="filepath"), outputs=gr.Text()).launch()