add new words

After creating model we found custom testing data were incorrectly paired so the final BLEU score was irrelevant. I suppose custom testing data are independent on model and could be updated without the need of creating new model. To prevent any abuse updating testing data could be allowed only once a day. I really do not plan to invest $300 to fix this cosmetic issue. But now I can't compare the quality with other models reliably.

keine Idee sondern ein ätzendes unützes programm gerade offline stellt eurer translator app sich völlig nutzlos da. konnte kein deutsches sprachpaket mall fix installieren..ich habs mir schon gedacht mal ne app von microsoft probieren aber war wieder klar das eine gängelung eurer seits eingebaut ist..umsonst aber nutzlos

Is there a feedback loop for retraining when translations are incorrect?

How much text cleaning is necessary? Should we leave the punctuation in as-is or we should clean them? How much cleaning? Maybe a couple of examples would be good. This is regarding English and Japanese.

Sorry but I cannot realize what the point is if training is allowed in F0 plan but deploy and using the model is prohibited. May I know what can I do with just training the model? Thank you.

speed setting for speech like most other apps.

Our sales teams are requesting that our site and content be translated in Sinhalese. We would love seeing support for this language as part of the Translator API including custom translation options for integrating professional translation memories.

I'm working on a solution to convert between English and textual ASL (i.e. English with different grammar). I've organized the data but the translation pairs are extremely restrictive.

Please add bahasa Indonesia into translator of kindle

Custom Translator doesn't the language pair we used on Translator Hub
We recently migrated our translation project - English - Pashto, over to the Custom Translator. However, Custom Translator doesn't have this language pair as in the Hub. Can you please add this so we can create the necessary projects and proceed with our work?

Translation add a lot of Uppercase on translations.

Example :
From French :

"Fournit avec :
- Certificat de conformité
- Certificat d'immatriculation
- Rapport d'identification
- Certificat de visite aunCT
Dimensions d'encombrement : 240 cm x 470 cm x 310 cm"

To English :

"Provides with:
-Certificate of Conformity
-Registration Certificate
-Identification Report
-Inspection Certificate AunCT
Dimensions Dimensions: 240 cm x 470 cm x 310 cm"

Our problem is thes tranlation in english is used to translate in dutch.
And a custom translation dictionnary.
But the translation dictionnary is Case sensitive.
=> The system doesn't use our custom dictionnary.

Do you have any plan to support other language pairs in Custom Translator which do not include English, such as between German and French?

I have uploaded a dictionary and tried to translate particular phrase that's already in dictionary but it seems not working. It being translated overlooking the dictionary.

If I tries to lookup action of dictionary api, it always returns empty result. Any Idea around this?

Here is the lookup result: [{"normalizedSource":"信投国际有限公司","displaySource":"信投国际有限公司","translations":[]}]

After entering new project information (name, language paid, category etc.) and click "create" button. After this, page is routed and displaying last opened project data (category id, language pair, data/model tabs, upload buttons are active) - after ~10 seconds page gets updated to display the new project name, language pair and other info.

I guess during this period you are interacting with backend system to get new project GUID which is getting appended to the URL.

Instead of showing last project data, simply show the generic loading/project creation message and show actual project info once project GUID is available.

In Customer translator project grid, allow sorting so that list can be seen in sorted order.

As part of customization of NMT, I would like to request terminology translation using user-created glossary within NMT. We don't have enough # of translations to train/build custom NMT. Instead, we have 3,000+ terminologies used for NMT. I have already implemented dynamic dictionary feature with translator text APIs, but in some cases such as long sentences or complicated contexts, it does not work as expected. So, I would like to have a similar feature "Glossaries in Google translator toolkit" with MS translator APIs.

有很多专有名词是不需要也不应该被翻译的。这些词被翻译会对整个句子造成很大的影响。
翻译的时候是否应该可以让用户提供 一个 专有名词的列表,让这些词原样保留,而不被翻译. 或者是 对于专有名词有一定的格式,而让机器去智能去过滤处理!
在很多科技相关的文章中,保留专有名词不被翻译,会大大提高翻译后文章的质量和可读性.