Loquendo
Loquendo Text-To-Speech serving advanced multimodal mobile services for rescue forces - SHARE Project results

In December 2007 the final review meeting of the SHARE project - "Mobile support for rescue forces, integrating multiple modes of interaction" - took place in Brussels, establishing the successful conclusion of the three year EC FP6 project.

The SHARE project is an IST EU-project begun in November 2004 with the goal of developing an advanced mobile service system that provides critical multimodal communication support for emergency teams during rescue operations. To achieve this goal a consortium of 9 partners was set up: Fraunhofer IAIS (Project coordinator), Feuerwehr Dortmund, University of Paderborn C.I.K, Telisma S.A., Siemens AG, Aristotle University of Thessaloniki AIIA, National Center for Scientific Research "Demokritos", Tele Atlas, and Loquendo.

Within the duration of the project the consortium developed a very advanced prototype of a mobile information and communication system for large scale rescue operations. This prototype (SHARE system) is a multi-user system for on-site cooperation which supports the work of fire fighting organizations in the field. The rescue teams can benefit from mobile and bi-directional communication based on 3G (UMTS) infrastructure and mobile WLAN networks located on-site during the emergency operation; the SHARE system allows to exchange structured multimodal information resources, including audio, video, text, graphic and location information.

The architecture of the final prototype is the following:



The specific innovations developed by the SHARE project include:

  • A Mobile system architecture design that enables bi-directional, multimodal communication using Push-To-Share technology
  • Digital Message Form
  • Video sharing
  • Multimodal interaction
  • Robust speech and image processing under extreme conditions
  • Interactive digital maps with linked multimedia information
  • Structuring of required information using situation-dependent ontologies and multimedia data indexing capabilities

The value to workers within the rescue operation domain originates from three important benefits that outclass the tools used nowadays. These benefits are:

       1. the mobility and efficiency introduced to the work in the field
       2. the provision of reliable and global sources of information
       3. the greatly increased support gained from team cooperation

The services offered by the prototype replace analogue information and communication tools that are used at present for operation planning and management, such as paper maps, magnetic boards, paper message forms and walky-talkies. A digital operation map with interactive annotation functionality, a group-based audio communication system, a resource management tool and a text-messaging service are the basic modules needed for fulfillment of the typical tasks of a fire fighting officer in an operation.
Vocal technologies have a fundamental role in multimodal interaction when hands-free interaction is necessary; they are integrated in the client side of the SHARE system, in the Map Client application, the Retrieval application and in the Digital Message Form application.
Text-to-speech technology is used for reading text based outputs (i.e. the retrieved texts or the text messages exchanged in the Digital Message Form application), street names and map contents, and for alerts or information.

New algorithms were developed for Loquendo TTS in order to obtain a robust speech synthesis in noisy environments under extreme conditions, where intelligibility has to be guaranteed.
The main research activities concerned different aspects of text-to-speech technology.
First, pronunciation accuracy and speech fluency have been improved, especially for the voices and the languages used in the project.
Second, a new concatenation algorithm has been developed and integrated into Loquendo TTS: this algorithm allows smoothing away of acoustic discontinuities and pitch jumps that may hinder the synthetic speech comprehension.
Third, a new dynamic range controller module has been developed. This module is able to enhance speech perception; it allows increasing the perceived loudness of the signal and can be tuned so as to obtain the best results in different noise conditions. The improvements and results obtained with this algorithm have been presented at the Interspeech 2007 conference.
Finally, a parametric equalizer has been developed in order to allow control of the frequency response relating to the audio output of the equipment used and to obtain further improvements in intelligibility.

The SHARE project has been presented at many conferences and events, arousing interest in this complete and unique solution for the planning and execution of large-scale rescue operations.
The SHARE consortium has achieved the goals originally set for the project, thanks to the considerable skills and excellent level of cooperation among the project partners; the UE commission has rated the project as good to excellent.

On the SHARE web site (www.ist-share.org) it is possible to find further information about the project.