Create an ambient word counter
Use AI, Node-RED, MQTT and a connected IoT device to discover how fast you are speaking.
About this video
In the last few years, we’ve become surrounded by speaking, listening machines. This revolution was brought about by trying to teach computers to respond to sound the same way a person might, by modelling the functionality of the brain instead of trying to write code that defines how language can, and should, be used.
Computers now have an ever-growing capability to not only understand what words are being said to them, but the intent behind them too. We’ve gone from a fanciful idea to everyday reality in less than a decade, and these tools are now available to pretty much anybody with a computer and an internet connection.
Now that humans and machines can speak the same language, I thought it might be a good idea to create something that helped me communicate better. For as long as I can remember speaking, I’ve been told that I speak much too fast. That’s OK. I have no problem repeating myself. Sometimes I just get carried away. But when one’s job involves public speaking events, one needs to have a clear, precise diction to be best understood by the audience, which is easier said than done (no pun intended).
To help me remedy this “talking too fast” issue, I created a Node-RED flow that uses Watson Speech-to-Text APIs to transcribe the words I’m saying, count how many words are being said in a given time, and then pass a message along to an IoT connected device (the GlowOrb) with MQTT telling it to change color. If I’m speaking at a normal rate it’s green; Bit fast? It’s yellow; Way too fast? It’s blue (blue is bad). When the color changes, I know I need to adjust the tempo of my speech, rather than trying to watch a clock or look for clues in the audience as to how I’m doing.
This ambient word counter is just one example of how AI and IoT connected devices can be combined to help improve people’s lives and abilities. And, it’s fun to make! Node-RED makes it quick and easy to create this sort of solution with very little code. So, here’s a video tutorial that I filmed showing you how to make your own.