app.py

from transformers import pipeline
import gradio as gr

# Load model
pipe = pipeline(
    "automatic-speech-recognition",
    model="baaabaei/whisper-small-fa"
)

def transcribe(audio):
    text = pipe(audio)["text"]
    return text

# Create Gradio interface
interface = gr.Interface(
    fn=transcribe,
    inputs=gr.Audio(sources=["microphone"], type="filepath"),
    outputs="text",
    title="Persian Speech Recognition",
    description="Real-time Persian speech-to-text using fine-tuned Whisper-small",
    # examples=[["path_to_sample_audio.wav"]]  # Add sample audio if available
)

interface.launch()