Japanese Yoon (diphthong) (for example, キョ) cannot be recognized.
F
Fulltek.and
started a topic
almost 3 years ago
I have setup configuration lang to ja_JP in iink 1.5.1, and put assets(myscript-iink-recognition-text-ja_JP), but Japanese Yoon (diphthong) cannot be recognized. example: キョ ひょ ヒャ ミャ ミョ リョ
Best Answer
O
Olivier @MyScript
said
almost 3 years ago
Dear Fulltek,
thank you for contacting us and your question.
This is currently a limitation of the technology, that in the absence of more context (i.e. characters before and after) confuses the second character with small and big size. This is something we are working to improve, but cannot commit when it will be done.
Meanwhile, our recommendation is to avoid recognizing "isolated characters" but instead recognize sentences.
Also, as the "small character" often appears as second candidate, you can export as JIIX and look for the other word and character candidates, and propose these to the user, so that he can update the result accordingly.
This is currently a limitation of the technology, that in the absence of more context (i.e. characters before and after) confuses the second character with small and big size. This is something we are working to improve, but cannot commit when it will be done.
Meanwhile, our recommendation is to avoid recognizing "isolated characters" but instead recognize sentences.
Also, as the "small character" often appears as second candidate, you can export as JIIX and look for the other word and character candidates, and propose these to the user, so that he can update the result accordingly.
In the later, we have worked on the recognition of the Japanese Yoon, and recognition should have been significantly increased. We then invite you to give a try.
Fulltek.and
I have setup configuration lang to ja_JP in iink 1.5.1,
and put assets(myscript-iink-recognition-text-ja_JP),
but Japanese Yoon (diphthong) cannot be recognized.
example:
キョ
ひょ
ヒャ
ミャ
ミョ
リョ
Dear Fulltek,
thank you for contacting us and your question.
This is currently a limitation of the technology, that in the absence of more context (i.e. characters before and after) confuses the second character with small and big size. This is something we are working to improve, but cannot commit when it will be done.
Meanwhile, our recommendation is to avoid recognizing "isolated characters" but instead recognize sentences.
Also, as the "small character" often appears as second candidate, you can export as JIIX and look for the other word and character candidates, and propose these to the user, so that he can update the result accordingly.
Best regards,
Olivier
- Oldest First
- Popular
- Newest First
Sorted by PopularOlivier @MyScript
Dear Fulltek,
thank you for contacting us and your question.
This is currently a limitation of the technology, that in the absence of more context (i.e. characters before and after) confuses the second character with small and big size. This is something we are working to improve, but cannot commit when it will be done.
Meanwhile, our recommendation is to avoid recognizing "isolated characters" but instead recognize sentences.
Also, as the "small character" often appears as second candidate, you can export as JIIX and look for the other word and character candidates, and propose these to the user, so that he can update the result accordingly.
Best regards,
Olivier
1 person likes this
Olivier @MyScript
Dear Fulltek,
to keep you updated:
We recently released the iink 2.0.1, that you can find on our github: https://github.com/MyScript
In the later, we have worked on the recognition of the Japanese Yoon, and recognition should have been significantly increased. We then invite you to give a try.
Waiting for you update!
Best regards,
Olivier