Generation of real-time lip transcription using 3D CNN and Bi-LSTM

Real-time lip transcription revolutionises communication accessibility by analysing lip movements to convert spoken language into text. The LipTrans model, utilising advanced neural network architecture and dropout regularisation techniques, streamlines transcription by directly mapping mouth movements to sentences. This technology enhances accessibility for the hearing impaired and improves automatic captioning accuracy for live events and online content. LipTrans achieves significant reductions in word error rate and character error rate, showcasing its increasing accuracy over training epochs. Its refined ability to interpret visual cues from lip movements marks a significant advancement in speech-to-text technology.