This work is certainly getting exciting and the word is out that we're demonstrating a professional set of deep learning capabilities in producing solutions for a wide range of business use cases! As data scientists, we understand the transferability of our skills. We know we can provide value by employing core skills when working on problems that we know are similar in structure but that may seem different at first glance. This couldn't be more true than in the next Deep Learning Project.  Next, we're (hypothetically) going to be working on a project in which a creative group has asked us to help to produce original content for movie scripts, song lyrics, and even music!  

How can we leverage our experience in solving problems for restaurant chains to such a different industry?  Let's explore what we know and what we're going to be asked to do. In past projects we demonstrated that we could take an image as input and output a class label (Chapter 2), we trained a model to take inputs of text and output sentiment classifications (Chapter 3), we built a NLP Pipeline for an open domain question answering chatbot where we took text as inputs and identified text in a corpus to present as the appropriate output (Chapter 4), and we expanded that chatbot functionality to be able to serve a restaurant in an automated ordering system (Chapter 5). 

In this chapter, we will implement a generative model which can generate content using the LSTM's, Variational autoencoders and GAN's. We will be implementing models for both text and images which can generate images and text for artists and various businesses.

What we'll learn in this chapter is:

1. Text generation with LSTM
2. Additional power of a Bi-directional LSTM for text generation
3. Deep (Multi-layer) LSTM to generate lyrics for a song
4. Deep (Mulit-layer) LSTM music generation for a song