A Comprehensive Review on Speech Synthesis Using Neural-Network Based Approaches

Speech is a primary mode of communication between human beings. With that, the need of creating artificial speech became a dream of humankind from the beginning of the 1980s. As a result of those experiments, many Speech synthesis systems were developed throughout the past few decades and further many advancements were made to the existing systems as well. However, with the beginning of the Artificial Intelligence era, these advancements were rapidly improved with the aid of neural networks. These newly found advancement techniques and methods were led to reemerge the research area of "Speech synthesis using neural networks-based approaches". This paper reviews currently available speech synthesis techniques, techniques that use neural networks-based approaches, advantages, disadvantages, and limitations. Further this review paper intends to suggest a new hybrid approach and what are modifications can be done in the upcoming future as well.