Developing Audio Narratives from Visual images for the Visually Impaired

The proposed model integrates voice commands with image processing and caption generation techniques to enhance accessibility, particularly for visually impaired individuals. The system begins by using a microphone input, where the user’s voice is processed to select an image from a predefined folder. Once the image is selected, a combination of Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks is employed to generate a descriptive caption for the image. CNN is used for extracting meaningful features from the image, while LSTM is responsible for generating a coherent and contextually accurate caption based on these features. After the caption is generated, the system converts the text into voice output, allowing users to receive the image description audibly. This voice-based image caption generation system provides an intuitive interface for users who may not be able to visually interpret images, enabling them to navigate and understand visual content through voice. The integration of voice input for image selection and voice output for caption delivery makes the system particularly user-friendly and accessible, promoting the use of image processing in assistive technologies. This model demonstrates how combining image processing with Natural Language Processing (NLP) can create impactful solutions for real-world accessibility challenges.

Keywords: NLP (natural language processing), CNN (Convolutional neural network), LSTM (Long short-term memory), RNN (recurrent neural network)