Training Sound Model on TensorFlow and Launching Prediction on Android

In this research paper, the topic "Training Sound Model on TensorFlow and Launching Prediction on Android" is presented. The purpose of the research is to study the algorithms that are part of the neural network libraries, implement them in a mobile application using Android OS platform, and make computations of the accuracy of built model to make a conclusion of the quality of finished project. The main problem described and solved by developed application is the classification of sounds using the trained neural network using a model based on Tensorflow library. The observed objects are sound recordings actually being the digital files containing the levels of the waveform signal. The subjects of the study are algorithms for distinction of different traits inside a sound signal among the continuous thread of frequency levels. The work consists of four main sections: introduction, which includes problem statement, background of the study and professional significance, the main body that consists of the literature review, methodology and the anticipated results sections, conclusion and the list of references.