Applications are innumerable, such as, each time security is involved (for instance: selective fire alarm giving accurate information about an event, precise information to the dispatching in an industrial process, etc) and also when 'smart interfaces' are required (for example: friendly access control, information/tourism kiosks, vending machines, ticket dispensers, robotics, etc). Our multilingual approach will allow voice messages to be delivered in a language chosen by the user or defined for him (for example: case of an access card containing a language parameter). This possibility is very useful in multilingual organizations, areas or countries. At present, we have the TTS technology for the following languages: German, Spanish, Italian, French, British and US English, Dutch, Portuguese and Brazilian Portuguese. Other languages can be added, upon request, in the future, as long as only TTS is needed.
Thanks to the multilingual WinDi technology described in this WEB site, the speech interfaces that will be used inside any 'industrial process' will have a certain linguistic independance from the host computer (or any programmable logic controller). The available WinDi technology is able to act by itself in a predefined context, which noticeably simplifies the integration of a voice interface on the host side. This allows us to take care of the complete multilingual text-to-speech design. For example, let's assume that the speech interface received an English text message that has to be read aloud; this request could contain various parameters as conjugation tense, interrogative/negative mode and of course the target language.
As described above, the possibility of adding linguistic parameters allows the system to be very open while providing text-to-speech with a powerful multilingualism.
We can develop such projects upon request and according to your requirements. We suggest you to contact us (Tel: +32-10-454.226; Fax: +32-10-454.995 or E-Mail: firstname.lastname@example.org) and talk about your project. Please read below the explanation about an application in the TTS context.
This project will serve as a voice interface for any computer having at least an RS-232 port and that should be able to provide multilingual voice information.
In order to do this, we have developed an interface program using one RS-232 port inside a PC. The requests received must follow a defined syntax to instruct the program properly. When a request is accepted, our Voice for WinDi program is activated to turn the text into speech. This communication program supports the XON-XOFF protocol and/or the STX-EOT one as the main protocol (its selection is automatic and host-driven). A whole series of feedbacks (can be parametrized) are returned to the host computer in order to inform it about the process status. The exchanges security has been extended to be reliable (all requests can end with a CRC). The ACK-NAK function is included. We have added a DATASCOPE feature into this program that displays all exchanges between the host and the program. This possibility is very convenient for maintenance purposes. For example, the 'xon' or 'xoff' characters are flagged on screen as '[XON]' or '[XOFF]' which is much clearer than a cabalistic sign. To follow is the program screen:
What is required to run this version of our Multilingual Speech Producer?
Any kind of PC running Windows 9x with 32 Mbytes RAM, a sound card and a hard disk (or Silicon Disk Drive). There is no need for VGA screen, floppy disk/CD-ROM drive, mouse and keyboard (except during installation/configuration; BIOS and Windows must be parametrized to boot without these units). There are a lot of solutions available on the market and we managed to produce the required hardware and software at the unit price of +-$500 for a quantity of 10 (this price includes a Cyrix MediaGX motherboard, small PC-AT case, 2 Gbytes HDD, 32 MBytes RAM, Windows 9x, TTS and WinDi licenses!). This means that such a solution is affordable and can be implemented in all kinds of environments. The following picture shows our own design:
Back To Home Page