multimodal berry, mmberry

vocal applications library, vocal application tools, multimodal sdk

skip menu

simple and natural:

use your voice

VoiceXML voice applications and multimodal applications

Multimodal interaction extends the potential of traditional applications.

If you require any information, you can listen to it and see it on your telephone.
Communicate with any management system by filling in forms and moving through menus.
Speak with the navigator to request information on the path and on services.
Use your voice to control your house.

Architecture

MMBerry layout click to zoom

How it works

MMBerry consists of a series of dll and header files (.h). The dlls contain the implementation of all the logic of the voice technologies used and the VuiManager.

The header file makes the VuiManager interaction interface available to the programmer.

The structure of the library has been designed so as to hide from the end user all the details relevant to the implementation.
Indeed, the header file only shows standard data types and makes no reference to low level objects.

Thanks to MMBerry, it is possible to write multimodal applications in C++, Java and .NET. In fact, there exists a complete implementation of MMBerry for Java and one for the .NET environment.

MultiModalBerry

MMBerry is a framework with which it is possible to develop multimodal applications rapidly and without having to know the low-level details of the underlying technologies.

MMBerry allows for the development of multimodal applications independently of the engines.

With MMBerry, it is possible to transform a classic application into a multimodal application by writing just a few lines of code.

Structure

MMBerry is made up of:

  • a Core library, structured from the following parts:
    • Voice Command Gateway: to write multimodal applications in C++, including voice recognition and synthesis functionality;
    • log file generation;
    • catch events from TTS and ASR engines;
  • JNI interface for writing Java applications with MMBerry in native mode.
  • .NET interface for writing applications with any framework language.
  • VAD (Voice Activity Detection) to optimise the use of the ASR motor in continual mode without being a burden on the recogniser itself.
  • Audio Dump allows for recording of the audio buffer which is sent to the ASR. This is especially recommended for debugging and application tuning.
  • Advanced logging allowing for the highlighting of each single event within the library and, above all, to have control of the recognition events (grammar loaded, utterance, recognition, confidence, etc).

The following tools are also supplied together with the libraries:

  • Voice recognition grammar editor.
  • Log analysis tools, which allow for analysis in real-time and in detail of the log files generated by an MMBerry application.