app.py

import whisper
import config
import gradio as gr


def get_text(audio_path):
    model = whisper.load_model(name='base', download_root=config.MODEL_DIR)
    results = model.transcribe(audio_path)
    return results['text']


label = gr.components.Text()
vd = gr.components.Audio(type="filepath")


iface = gr.Interface(fn=get_text, inputs=vd, outputs=label)
iface.launch()