app.py

import gradio as gr

import whisperx
audio_model = whisperx.load_model('tiny', 'cpu', compute_type="float32")

def generate_answers(audio_q):
    trans = audio_model.transcribe(audio_q)
    audio_res = ""
    for seg in trans['segments']:
        audio_res += seg['text']
    audio_res = audio_res.strip()
    return audio_res

with gr.Blocks() as demo:

    gr.Markdown(
    """# Testing wisper
    """
    )

    # app GUI
    with gr.Row():
        with gr.Column():
            audio_q = gr.Audio(label="Audio Question", value=None, sources=['microphone', 'upload'], type='filepath',show_download_button=True)
    with gr.Row():
        answer   = gr.Text(label ='Answer')
    with gr.Row():
        submit = gr.Button("Submit")
        submit.click(generate_answers, inputs=[audio_q], outputs=[answer])
        clear_btn = gr.ClearButton([audio_q, answer])
    
if __name__ == "__main__":
    
    demo.launch(share=True)