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, actual_transcription):
    text = pipe(audio)
    return text['text']


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

iface = gr.Interface(
    fn=transcribe,
    inputs=[
        gr.Audio(source="microphone", type="filepath"),
        gr.Textbox(label="Actual Transcription", optional=True)  
    ],
    outputs="text",
    title="arabeasy",
    description="Realtime demo for Lebanese Arabizi speech recognition",
    allow_flagging='manual',  # Enable manual flagging
    flagging_callback=hf_writer
)

iface.launch(share=True)