Speech recognition has made ground-breaking progress in the last few years, thanks to the growing availability of faster hardware, bigger data, and better algorithms. Today real-time, near human-level speech recognition is available on desktop and mobile devices, and leveraging this technology to build your own speech-enabled applications is easier than ever before. Many open source tools for speech, including state-of-the-art libraries like CMUSphinx and kaldi offer powerful APIs for developers to perform live speech recognition. In this talk we will show you how to train a language model and build a custom voice-control interface from scratch using open source tools and real-world data. We will demonstrate a plugin built at JetBrains for controlling your IDE by voice, and give you advice for implementing your own speech applications. No prior experience is necessary.