Virtual speakers characters are the result of a research program within Cube systems, largely supported by the ANVAR.

Assistants have been used since 2005 to produce short broadcasts for TV channels and Neoscreen They act as a speaker, presenting news, programs and automated games on our channels.

As they are so easy to user, and on screen performance is excellent, program editors use them a lot. The remove the need to produce filmed broadcasts, which are much more costly in terms of materials, manpower and time.

Their use on web services naturally derives from these applications.

Based on a Flash program, assistants can be easily integrated into any web application They enrich our interactive products such as KWIZ, or merely provide a voice for KP.

Multilingual characters read out simple texts thanks to voice synthesis tools But they can also react to natural recorded voices In this case, lip synchronisation is ensured by voice recognition tools which identify phonemes..

Assistants can interact with a multimedia décor and present various multi-media subjects, perfectly synchronised.

There are multiple uses:

- Virtual speaker or presenter
- Virtual guide or hostess
- Information agent
- Virtual advisor
- Virtual professor
- Vendeur virtuel
- Virtual vendor

The Virtual speakers program has ambitious goals:

1° To create a conversational agent to provide immediate assistance for users of an electronic service In a specific language and within a limited context the user speaks to the machine and gets responses to the question asked from the agent.

2° To enable the agent to provide the user with information partly in the form of voice synthesis (indications) and in the form of multi media documents: display of a page on the screen, printout of the documents requested.

3° To allow the agent to interpret a voice request (made if possible in natural language), to associate it with a subject available in its knowledge base and to respond to it.

4° To enable the visual representation of the agent (in the form of a speaking character reacting to a text).

5° To expand the knowledge base of agents to react to different languages.

The scope of application is broad, but three areas for research have been focussed on, as they appear to have an immediate economic result:

- the service for the production of short broadcasts (news, games) for new television channels, web services, WebTV

- the public sector to improve services for people in difficulty in various situations (requests for information, obtention of documents etc.)

- the medical sector (services for people in care centres) and help with home care