Does the presence of other language text slows the machine translation speed of SDL?

Hello SDL Community Team, 

I have this query in mind, as to whether the presence of 2 different language texts, for example English and Bhasha in an MS Excel column, can be the reason of slowing down of the speed/performance of SDL MTC tool? If not what could be the underlying reason for this slowness in speed. Is it due character limit per translatable file? Please shed some light here

Please revert if you need any further information.

Thank you in advance,

Regards,

Manashi

  • Hello Manashi,

    How are you translating XLSX documents ?  Are you uploading it to SDL Machine Translation Cloud or using a plugin from SDL Trados Studio ? 

    Uploading it to MT Cloud is the fastest solution. Parsing and translation happens directly in MT Cloud.

    Using API from SDL Trados Studio is slower, since the plugin sends API calls over HTTPs in batches of strings and transfers results back before a new batch is sent. This method is suitable for smaller documents or string by string translations, but not 100.000+ word documents. 

    Without seeing the MS Excel document, it is hard to tell. You don't say  if your bilingual file has a column with target language hidden or not. How large it is in KB/MB.  How many words it has ? 

    Parsers will try to extract all unhidden text and try to perform translation on all words, so if there a text you don't want to be translated, please hide it.

    Some Excel documents may  contain unnecessary formatting applied to the whole column ( e.g applied 100.000+ rows) which may result in parser scanning rows/columns with the formatting unnecessarily.    - try to copy and paste selected text only to a new Excel document. - the size of Excel document will be reduced visibly. 

    Please note that very large documents with 100.000+ characters may be queued in the system for longer than if particular LP is busy with other requests from other users.

    Badly formatted text with tags, very long sentences not segmented correctly , many unknown source words will add up to the processing time. 

    e.g.  if you have HTML content pasted in cells and you upload it directly to  SDL Machine Translation Cloud, translation will be slower of lesser quality and will use more words of your subscription. This is because the content in cells is parsed as plain text.

    You can still raise a case with SDL Support and provide a sample document for further investigation

    Kind Regards

    Dusan