"Currently we are able to translate at the text level, but the text that can be translated is extremely limited. The technology must convert the Japanese language that is input into a string of sign language words.
We use samples that have the same content in both Japanese and sign language, compare them to what is actually input, and then replace words that differ to achieve the translation. Meanwhile, to create the sign language CG we need to make automatic transitions between words, and to a degree we are able to create that. However, there are parts that we are not able to express well yet, and we have made an interface in which a human can correct those parts."