Audio for Visually impaired using Inception V3 and LSTM

Humans depend heavily on their sense of sight, and its loss can have a major negative influence on a person's quality of life. Being blind can make it more difficult for a person to carry out daily activities on their own, such as reading, identifying things, and navigating new surroundings. The WHO estimates that there are about 285 million visually impaired people in the globe, 39 million of whom are blind. The concepts of picture captioning and TTS, which uses deep learning to transform an image into text and audio description, are combined in Visual Audio for Visionless. This process is difficult since the system must first comprehend the image's content before figuring out how to convert it into natural language captions that accurately convey the image's meaning both orally and in writing. LSTM with attention functions as a decoder and creates the word-by-word caption in the suggested system, conditioning the prediction on the previously generated caption, while CNN (Inception V3) acts as an encoder. Because of their complementary strengths, Inception V3 and LSTM, which are utilized for picture captioning, can produce good results. In the end, as soon as the system records the situation as an image, captions and audio descriptions are produced for the photographs in real-time.