The first thing to understand is what a license gives you. To do that we must first understand a bit about how your speech applications communicate with the LumenVox Engine.
So you have an application, callers call into it and you would like to know what they say. In order to do that the application will open what we refer to as a Speech Port to the Engine. A speech port is simply a connection between computer applications. So the port is opened and the audio of what the caller said is sent into it. The Engine works on the audio, decodes it and gives the audio back to the application. At that point the application will close the speech port and uses the results to send the caller on his way.
For every time that you have a speech port open, you're going to need one license. That license is freed up when the port is closed. Another way of thinking about it is, for each simultaneous speech recognition you're going to need one license. If you have five callers talking to the application at one time and you want them all understood at the same time, five speech licenses would be required.
The question we get all the time is, "How many licenses will I need?" The answer is, it depends. It depends on your specific application requirements:
A good idea is to contact our sales department and they can walk you through your application requirements and give you a general idea of how many licenses you'll probably need.
The next thing to talk about is the two license types LumenVox offers. We offer a Lite License and a Full License. The way that these two licenses differ is the number grammar entries that you can have decoded at once. A grammar entry is simply a word that you want recognized. For example, if you ask a caller a simple question such as "Would you like fries with that?" The caller can either answer "yes" or "no." That's just two words to be recognized.
The Lite license allows up to 500 grammar entries at a given prompt. In this case using the yes/no example, the Lite License would be fine because there are only two grammar entries.
If you have a more complicated application, such as a call router for 500 employees, you might think that the Lite License would work. Since the Lite License provides 500 grammar entries and there are 500 employees this should work fine. Well this would not be a good idea because each one of those employees probably has at least a first and a last name, and also alternate pronunciations and nick names may be required. Each one of these would count as a grammar entry. For example, the name Stephen, you may also need to include Steve in the grammar, and also this spelling is with a "p," "h," so people may pronounce it "Stefan" if they've seen it written. This would be three separate pronunciations for the first name alone. So if you had 500 employees, you could probably expect to have 1,000 to 1,500 grammar entries.
So in that case you would require our Full License. This license provides up to 12,000 grammar entries at a time.
Another frequently asked question is, "For my application do I need a Full or a Lite license?" The answer again is that it depends more on the application so contact the LumenVox sales department for more information.
Another term to understand is the concept of a deployment:
The last concept to address is the License Pool, and how it relates to our general idea in the way we would work in a distributed environment. If you chose to you could put your speech application, the speech engine and the license server all on the same box. However, that is probably not the best idea because, as mentioned earlier, licenses are not easily moved. So, let's say you're in development, and you want to test your application in multiple environments. You want to use different hardware types, different environment variables, etc. But you can't easily move your licenses. So the solution is to put all your licenses on a license server, and put your engines on different machines for your testing. This also works very well in a production environment, because you'll probably want to have a large load of calls split up among multiple machines. Again, the way to do this is to have a single license server, and a bunch of different engines running on different machines. LumenVox supports load balancing and working in a very distributed manner. In order to provide an example of how this works we'll illustrate the concept of the license pool. So you have your license server and it has a pool of licenses, let's say there's five licenses available.
So as soon as a Speech Engine running on a different machine needs a license for a decode, it will go over the network and ask the license server if it has a license available. The server will check it's pool and see that it has five licenses, it will reserve one for the request leaving four available in that pool. The license server would then send the speech engine a message saying that you have a license and you are clear to do your decode. If at the same time a second speech engine also needed a license, it would go to the license server, the server would once again check it's pool, and see that there are four available. It would reserve a license and send a message to the second speech engine letting it know that it could do its decode. When that first speech engine finishes it's decode, it would tell the license server that it is done with the decode and that the server can have the license back. The license would be put back into the pool which means that there would once again be four licenses available. If there ever a case where the license server is out of licenses in it's pool and a speech engine asks for a license, the server will say sorry but you'll have to wait until a license becomes available. This is good because if you have load balancing distributed needs we work very well in that environment. You will not have to worry about getting licenses on every machine and attempt to guess at how many calls a specific machine will get.
So let's review some of the concepts that we have discussed:
In the next video we'll introduce our Web-based Interface to you. This will allow you to:
© 2016 LumenVox, LLC. All rights reserved.