Advances in Speech and Text Processing for Dravidian Language: A Comprehensive Review

This review explores recent advancements in speech and text processing for local and under-resourced languages, emphasizing Dravidian languages. This work analyze 40 studies on topics such as speech recognition, sentiment analysis, emotion detection, and language identification, highlighting the growing use of deep learning techniques like CNNs, LSTMs, and transformers. Trends include multimodal and code-mixed language processing, and the use of both traditional and modern feature extraction methods. Key findings emphasize the success of hybrid models, the value of transfer learning and the need for high-quality datasets. The review also identifies gaps, including handling dialectal variations and addressing ethical considerations in language technology and also Industry, innovation and Infrastructure.