Znakovko
Znakovko is a text to sign language conversion tool that aims to ease communication with the deaf and hard of hearing. The tool was made in cooperation with the DIPteam. It first converts the input text into a sequence of symbols which it then translates into a video sequence of signs. The sign vocabulary with which the tool works is 200, and it achieves accuracy of 95.51% at the level of symbols.
The problem addressed by the tool is the difficulty of following written text for hearing impaired people. This comes from the lack of knowledge of spoken language and how letters sound. Moreover, sign language works differently from spoken language - in it, thoughts are expressed through sequences of symbols that do not have grammatical distinctions such as gender, number, person, and tense. A few examples are given in the following table:
| vocal | sign |
|---|---|
| Snezhe knew someone was waiting for her there. | Snezhe knows someone waiting there |
| The family of sparrows lived on the most beautiful tree in the neighborhood. | family bird lives beautiful tree place |
| Slave and his father loved to help animals. | Slave his father very much wants help animal |
Znakovko is described in the paper:
- Spasovski, Stefan, Branislav Gerazov, Risto Chavdarov, Viktorija Smilevska, Aneta Crvenkovska, Tomislav Kartalov, Zoran Ivanovski, and Toni Bachvarovski. “Towards a System for Converting Text to Sign Language in Macedonian,” ETAI 2021, Struga, Macedonia, Sep 23 - 24, 2021. preprint
Acknowledgement
The work on Znakovko was supported by Victoria Smilevska and Aneta Crvenkovska with their expertise in sign language. The video recordings of the signs were made with the help of Gordana Stamatoska.