app.py

import gradio as gr
from transformers import pipeline
import os


pipe = pipeline(task="automatic-speech-recognition", model="geokanaan/Whisper_Base_Lebanese_Arabizi")
def transcribe(audio):
    sr, y = audio
    y = y.astype(np.float32)
    y /= np.max(np.abs(y))

    return pipe({"sampling_rate": sr, "raw": y})["text"]

HF_TOKEN = os.getenv('WRITE')
#hf_writer = gr.HuggingFaceDatasetSaver(HF_TOKEN, "geokanaan/flagged_Audio_Lebanese")

iface = gr.Interface(
    fn=transcribe,
    inputs=[
        gr.Audio(sources="microphone")
    ],
    outputs="text",
    title="arabeasy",
    description="Realtime demo for Lebanese Arabizi speech recognition",
    #allow_flagging='manual',  
    #flagging_callback=hf_writer
)
iface.launch()