Putting the user in control
The Intelligent Speech Interfaces™ platform from Voxi focuses on the unique features of speech: its conciseness and its ability to exactly express the speaker's intentions.

Applications built with the patent-pending technologies of the ISI-platform let users talk in natural sentences. The ISI-platform also employs AI-like reasoning so it can fulfill high-level user requests by breaking them down into a chain of simple executable tasks.

The applications are modeled using an object-oriented approach that enables the platform to handle linguistic elements using object-oriented methods. This makes it possible to both understand and act on the spoken words, as well as to produce context-to-speech feedback to the user.

Platform overview
The platform consists of three major parts:
- The Speech Recognizer
- The Natural Language Engine
- The Sound Renderer
The speech recognizer transforms sound into text. It uses application context information to improve recognition.

The natural language engine translates the linguistic information from the recognizer into elements of the application's object model. When the request has been analyzed, the proper action is initiated. The language engine can be connected to multiple applications at the same time, providing for seamless user interaction with all the applications.

The feedback to the user can come from both the language engine and the application. It is mapped through the sound renderer which can be freely configured to process the sound before being played to the user.
Technical white paper
To request our Technical Platform White Paper, please contact us at techinfo@voxi.se.