Understanding Speech Technology
In deploying an effective speech recognition application, there are many important
details to be considered. The following are a list of terms commonly
used to describe the technology, including links to further
information.
Barge-In
Barge in is the ability to interrupt the speech application with a
response. Callers who use the system more frequently can avoid
listening to their options and simply state what it is they are
looking for. At the same time, those who are less familiar with the
speech application will be provided with "helper prompts," to aid them
in their transaction.
Grammar
Grammar in a speech application includes all the recognized phrases
for a particular response. For example, a question seeking a yes or no
answer, may also recognize when the caller responds with variations of
these responses: yeah, yup, nah, nope... etc. It will also recognize
natural speaking, such as the caller saying "ummm, yes." The
application has a number of responses it looks for, and it listens to
the caller for those responses. Our applications handle many different
dialects, and recognize grammar specific to the regions where they are
implemented.
Language
Options
An additional feature of speech is the recognition of multiple
languages. Speech applications can run concurrently in several
language options including American English, American Spanish, French
Canadian and more. VoicePort offers over 140 language options.
Tuning
Tuning is the process of customizing a speech application to your
particular needs. No two people say the same word exactly the
same. Tuning can include editing prompts and grammar to reflect
regional dialects in order to improve the automation rates of your
application. VoicePort engineers listen to recorded phone calls to
hear where the errors occur, and make any necessary adjustments for
better performance. A great deal of tuning occurs before
offical deployment, but it is an on going process to continuously
improve automation rates.
Voice User Interface
The Voice User Interface (VUI) is the most important aspect of any
speech application. It is what differentiates a good application from
a bad application. VUI's must be designed conversationally, and prompt
the caller for logical responses. When the prompt is not enough, the
VUI must provide the caller with additional support to guide them
through their call successfully. When a caller fails to respond how
the system expects, it is not the caller who has failed, it is the
speech application. VoicePort's dynamic VUIs utilize a continuous
improvement process. Our goal is to create the most efficient
application possible, meeting both you and your customers' needs.
VXML
VoiceXML or VXML is the industry standard programming language for
speech applications. VoicePort applications operate on a
Genesys VXML platform, the leader in voice platform providers.
