VisionVoice/accessibility-assistant

1# -*- coding: utf-8 -*-2"""Assignment 4_Visual Assistance_with TTS.ipynb3 4Automatically generated by Colab.5 6Original file is located at7    https://colab.research.google.com/drive/1LBTUSQuENvlt7NtPgp_-3Rd0SLIxQooR8"""9 10import gradio as gr11from PIL import Image12from transformers import BlipProcessor, BlipForConditionalGeneration13from gtts import gTTS14 15# Load BLIP Vision-Language Model16processor = BlipProcessor.from_pretrained(17    "Salesforce/blip-image-captioning-base"18)19 20model = BlipForConditionalGeneration.from_pretrained(21    "Salesforce/blip-image-captioning-base"22)23 24# Main function: Image → Text → Speech25def describe_image(image):26 27    # Preprocess image28    inputs = processor(image, return_tensors="pt")29 30    # Generate caption31    output = model.generate(**inputs, max_new_tokens=50)32 33    caption = processor.decode(34        output[0],35        skip_special_tokens=True36    )37 38    # Convert text to speech using gTTS39    tts = gTTS(text=caption, lang="en")40 41    audio_path = "output.mp3"42    tts.save(audio_path)43 44    # Return both caption + audio file45    return caption, audio_path46 47 48# Gradio Interface (Text + Audio output)49demo = gr.Interface(50    fn=describe_image,51    inputs=gr.Image(type="pil"),52    outputs=[53        gr.Textbox(label="Image Description"),54        gr.Audio(label="Audio Description")55    ],56    title="VisionVoice Accessibility Assistant",57    description=(58        "Upload an image and receive an AI-generated description "59        "with text and spoken audio output to support accessibility."60    )61)62 63demo.launch()