app.py

from transformers import pipeline
import gradio as gr

p = pipeline("automatic-speech-recognition", model="aware-ai/wav2vec2-base-german")

def transcribe(audio):
    return p(audio, chunk_length_s=10, stride_length_s=(4, 2))["text"]

gr.Interface(
    transcribe, 
    [gr.Audio(source="microphone", type="filepath")],
    
    [gr.Textbox()],
    live=True
    ).launch(server_name = "0.0.0.0")