app.py

import gradio as gr
from transformers import pipeline

# initialize ASR pipeline
asr = pipeline(
    task="automatic-speech-recognition",
    model="vhdm/whisper-large-fa-v1",
    device=-1  # CPU; set device=0 for GPU
)

def transcribe(audio_file):
    """
    audio_file: path to WAV file (Gradio mic or upload)
    """
    if not audio_file:
        return "No audio input detected."

    try:
        # run ASR
        result = asr(audio_file, chunk_length_s=30, stride_length_s=[5,5])
    except Exception as e:
        return f"ASR error: {e}"

    text = result.get("text", "")
    return text

# gradio interface
iface = gr.Interface(
    fn=transcribe,
    inputs=gr.Audio(type="filepath", label="Record or upload audio"),
    outputs="text",
    title="Persian ASR",
    description=""" Speak in Persian or upload an audio file."""
)

if __name__ == "__main__":
    iface.launch()