Join "Natural Language Understanding with Python, Neural Networks-TensorFlow and ROS Integration" on Robociti to improve your Speech Processing skills and learn how to create a Natural Language Understanding (NLU) module that can act as the final stage in a Speech Recognition pipeline on a live or pre-recorded audio stream, matching recognized words or sentences to a set of expected outcomes for use in a specific application. To achieve this, popular frameworks like Python programming for dataset creation, TensorFlow and Keras for Neural Network model construction and fast deployment, and ROS (Robot Operating System) for further integration with other modules in larger AI & Robotics projects are utilized. Firstly, the process of creating a dataset of the expected outcomes based on a specific application is presented, along with any data processing required. Then, the method for using the dataset constructed to train a suggested Neural Network Model based on TensorFlow and Keras is analyzed, along with the way to calculate the accuracy of the model produced and tips about how to optimize the training process. Finally, the module is wrapped in the ROS framework for the integration in a larger Speech Recognition application and students work on improving the final result.