StoryGraph is a speech to image application. It's main use case is for bed time story telling: imagine telling a story to someone and having an app render a number of images for each phrase you speak.
We use Watson Speech API for speech to text transformation. After that we feed each phrase into Google's Syntax Net to adnotate it with POS. Then we apply a model to summarize the phrase to a number of search terms that we use to search images using bing image search.