Text to speech synthesis software Raspberry Pi

******The goal of the project is to run on resource constrained devices***** A text to speech synthesis solution for English (American and British) utilizing neural networks that can be deployed on small devices. This solution will be utilized on devices like the Raspberry Pi size and smaller. It is assumed that the training of the model will occur off line and the text to speech synthesis will happen on board the device without interaction with a data center, cloud, or any other device except when loading or updating the model. The solution provided must included specifications for the expected speed and accuracy performance as well as the training resource requirements such hardware and software and time resources. The proposed solution must include: - Functionality o How Generality is to support multiple languages and what those languages are o Complexity of the language supported o Number of words supported o Dialects or accents supported - Performance o Speed o Accuracy - Training Requirements o Hardware/Software Resources o Time Skills: AI Speech Synthesis Natural Language Understanding (NLU) Languages: Python and C++ Frameworks/Libraries: TensorFlow, PyTorch, We plan to test the software on the Raspberry Pi 3 Model B+