I use my Alexa Echo devices daily for various purposes. While I find it incredibly useful, Alexa does have limitations, and it would be great if it were more customizable. I decided to try making my own voice assistant that I can fully customize to my needs. I attempted this many years ago when I was first learning how to code, and mostly failed. At the time, the only method available to me for speech synthesis was to have a database of pre-recorded audio files that I could call when appropriate, and that was very limiting. Now however, there is a nifty text-to-speech conversion Python library called “pyttsx3“. Using this library alongside Google’s cloud speech recognition API (which can be accessed using the SpeechRecognition 3.10.0 library) allowed me to put together a very capable voice assistant.
At the moment, the voice assistant only has a few capabilities. It listens for a wake word to activate, and then can give you the current time or date, it can set a timer, it can search Wikipedia to answer questions, and it has some small-talk options. I intend to build in many more features, including fetching news headlines, giving weather forecasts, web scrapping various sites, setting reminders, sending emails, and controlling some of my smart home gadgets.
Some of these features are going to require a web server running some custom JavaScript that can gather data from online sources and send that data to my voice assistant via JSON requests. So building the web app that will serve as an API to my voice assistant is one of my next steps.
I also want the voice assistant to run on hardware as an IoT device, so that is going to take some tinkering. I have ordered the components I need to build this, and I am waiting for them to arrive. The final step will be deciding what to use to house my project. Maybe 3D printing an animatronic, talking head? Or recreating the Hal 9000? I haven’t decided yet.