Speaker 1: Hi, and welcome to The Shed, where I get to show you new ways to use Speechmatics transcription in some real-world scenarios. Hopefully these videos will inspire you to come up with some new and interesting ways of using our APIs. In this episode, I'm going to show you how I use our Realtime API to create an agent assist tool that could be helpful for agents who work in call centers. I'm not a developer myself, but I've written this very simple UI in React just to show you how you could use a real-time transcription. And in this case, what it's going to have on the left is the transcription, and it will then be tracking keywords that I've defined earlier on. And it will then detect those keywords and look up some data in a knowledge base, and it would then display that as a battle card. The UI allows me to start and stop a call. What that does is it opens up the microphone with the browser and starts sending through transcription into our engine, and then you'll see you start to get transcription on the left here. S1 is Speaker 1. If there was a second speaker on the call, then there would be also S2. You can use that to drive other downstream aspects of your workflow. So as we're transcribing, I want to demonstrate this as if it was a salesperson at Speechmatics. One of the common things we have is the question of, well, how well do you perform? What's your accuracy? And in this case, what's the word error rate? And so maybe a customer say, hey, how do you compare to Google? And as it detects the word Google, we'll get a battle card come up, and it will show us the word error rates here for Speechmatics against Google. And I can change here the language as well. So I can see it's Spanish, and I can also see in German. Now, if I was to speak about Amazon, for example, and how do they compare, we'll see that it updates the UI automatically to show me the data from Amazon. Also, we have a number of features around like speaker labeling and the other features we offer. But if a customer wants to ask us about diarization, a feature, again, we can bring up some relevant information just to help the customer and the agent know more about what's being spoken about. So this shows us the different keywords. We can close them in this UI, open up the other ones just to go back to ones we've previously checked, or we can actually see on the timeline previous keywords that were spoken. So this just gives you a bit of a flavor of how to do something relatively simple, but could be quite powerful in an application where you have a lot of complex information you want to surface just at the right time. You can clearly do more than just keyword spotting. You could do topic detection and you can do other things around like sentiment to detect if some caller is unhappy and you could provide a lot of different guidance for the agent to help them. So this has just been a very simple example of this. So I'm going to go ahead and stop the call. But thanks very much for joining and listening. I hope you found it interesting and I hope it gives you some ideas of what you can achieve with our technology. If you would like to make a video of your own for the Shed or showcase an idea you have yourself using Speechmatics, please do get in touch. We'd love to hear from you.