PhoneMyBot provides an advanced capability to optimize the results of the Speech-to-Text engines interpretation, by choosing the best engine vendor (among the ones it uses) and configuring the engine for the task at hand. To do this, PhoneMyBot needs to know the context of the conversation and only the chatbot can tell it what it is. So, PhoneMyBot supports an ever growing number of specialized contexts that the chatbot can set with the Conversation API the Call Actions ML, or pre-set using a regular expression rule. Follow these links to learn how to set a context.
Once the chatbot triggers a context, PhoneMyBot configures the Speech-to-Text engine for the next thing the user will say: the next utterance. For instance, suppose that the chatbot asks the user their car's license plate. Reasonably, what the user says next is the licence plate digits and letters.
You can also get familiar with this concept by watching this video: https://youtu.be/q9L0md4NwaA.
Using this capability from the chatbot can bring to much better results in utterance interpretation precision. Again, if PhoneMyBot knows that the user has been asked to provide a series of digits (an account number for instance), it can use an engine with better performance recognizing digits in a certain language. On a series of 10 digits, a 5% recognition difference on each one results in a 40% difference in recognizing the whole series.
Here is a list of the contexts currently available. They are per language and per country, since each country has its own set of needed information for common tasks. The only keyword you need to set the context is its name. This is in bold and between quotes in the following.
If you encounter a specific piece of information in your chatbot workflow that is not in the list and needs to be optimized, please let us know and we'll add it to the supported contexts.