IP Logo
distance keeper Computer Vision & Graphics  
Graphic Element West Graphic Element Middle Graphic Element East
 
Graphic Element Quadgray Start
Graphic Element Quadgray News
Graphic Element Quadgreen Organisation
  Image Communication
  Computer Vision & Graphics
  Immersive Media & 3D-Video
  Hardware Architectures & Implementations
  Embedded Systems
Graphic Element Quadgray Fields of Competence
Graphic Element Quadgray Fields of Application
Graphic Element Quadgray Alliances & Committees
Graphic Element Quadgray Products
Graphic Element Quadgray Events
Graphic Element Quadgray Staff
Graphic Element Quadgray Jobs
Graphic Element Quadgray Visitors
Graphic Element Quadgray Contact
Graphic Element Quadgray HHI Home
Group 2 Logo
   
 

Text2Video

The image-processing department at the Fraunhofer Institute for Telecommunications, Heinrich-Hertz-Institute has developed a system for the conversion of SMS messages into small animations which can be sent to the recipient via MMS. From the input text, speech is synthesized using a TTS (text-to-speech) system. Similarly, lip movements of an arbitrary real or synthetic person are artificially created synchronously to the spoken text. As a result, the chosen character reads the message to the user. Random eye and head movements, emoticons embedded into the text, virtual camera zooms or pans, and voice modifications are implemented in order to make the animation more lively. From the animation, a 3gp video is created that can be sent to mobile phones via MMS. Since the whole application works with exiting infrastructure (with an additional server at the provider), no software has to be downloaded to the mobile device and the animation can be viewed on any mobile phone with video support.


Applications for Text2Video

  • SMS to MMS conversion: An arbitrary character reads the text from an SMS
  • Individualized MMS commercials
  • Intelligent user interfaces: An avatar communicates with the user with user dependent information
  • Presentations on web sites

Please check the demo of the system


Features

  • Animation of synthetic avatars and real people
  • New characters can be created on demand from a single real picture of a person
  • Different TTS systems and phoneme alphabets supported (AT&T, Babel, ScanSoft)
  • Different languages supported (US/UK, German, French, Spanish, Polish, others on demand)
  • Support for emoticons to add facial expressions
  • Pitch shift for voice modification
  • Camera zoom and pan, eye and head movements for more sophisticated animations
  • Different encoding possible 3GP, H263, MPEG-4
  • No software required on mobile phone (except for video player)
  • Runs on a single Linux server with X and 3D graphics support

The software for the creation of these animations is available at HHI and can be licensed to customers. Other features and special extensions can be developed on demand.


Examples


Natural (left) and synthetic (middle) character reading a text. Right: Animation on a mobile phone.


Publications

J. Rurainsky and P. Eisert,
Text2Video: A SMS to MMS Conversion, Proc. 11. Dortmunder Fernsehseminar, Elektronische Medien: Systeme, Technologien, Anwendungen, Dortmund, Germany, pp. 41-52, September 2005.

J. Rurainsky and P. Eisert
Text2Video: Text-Driven Facial Animation using MPEG-4, Proc. Visual Computation and Image Processing (VCIP), Beijing, China, July 2005.

Flyer Text2Video



Contact

Dr. Peter Eisert
Email:
Phone: +49 30 31002 614
Fraunhofer Institute for Telecommunications
Einsteinufer 37
D-10587 Berlin
Germany