from transformers import pipeline
# Create Object Detection pipeline to Detect Objects in an Image
od_pipe = pipeline("object-detection", "facebook/detr-resnet-50")
#Create Text to Speech pipeline to Generate Audio Narration of an Image
tts_pipe = pipeline("text-to-speech",
                    model="kakao-enterprise/vits-ljs")
                    
#Build the Audio Powered AI Assistant
import os
import gradio as gr
from helper import summarize_predictions_natural_language  

def dictate_whatisin_image(pil_image):

    pipeline_output = od_pipe(pil_image)

    text = summarize_predictions_natural_language(pipeline_output)
    narrated_text = tts_pipe(text)
    (narrated_text["sampling_rate"], narrated_text["audio"][0] )
    return (narrated_text["sampling_rate"], narrated_text["audio"][0])
demo = gr.Interface(
fn=dictate_whatisin_image,
  inputs=gr.Image(label="Input image",
                  type="pil"),
   outputs=gr.Audio(label="Listen to what is inside the image", type="numpy",  autoplay=True),
   title="AI Powered Audio Assitant",
   description="Dictate what is inside the Image"
)
demo.launch()