In this chapter, we first gave a quick overview of speech recognition and how modern ASR systems were built using end-to-end deep learning methods. Then we covered how to train a TensorFlow model to recognize simple speech commands, and presented step-by-step tutorials on how to use the model in an Android app, as well as in both Objective-C- and Swift-based iOS apps. We also discussed how to fix a common model-loading error in iOS by finding out the missing TensorFlow op or kernel file, adding it, and rebuilding the TensorFlow iOS library.
ASR is for converting speech to text. In the next chapter, we'll explore another model that has text as the output, and the text there will be full, natural-language sentences instead of the simple commands in this chapter. We'll cover how to build a model to convert an image, our old friend, to text, and how to use the model...
