L&D Challenge: Translation

Amazon AWS, under their machine learning function, has some interesting tools that I’ve wanted to try out for a while. The two I decided to explore first are AWS’ Amazon Translate and Amazon Polly (a text to voice engine).

For this post, I'm going to explore: Amazon Translate.

The Amazon Translate tool intrigued me because it states that its neural machine translation system uses the whole sentence for context. It also allows you to upload custom terminology. I don’t know a single company that doesn’t have its own acronyms and lingo, and preloading that terminology can really help a translation project go smoothly.


The service also boasts of 55 languages and variants. The ones that stood out to me working in Texas were Mexican Spanish, Vietnamese, Korean, Arabic, and German.


I do love a good deal, and this service is right in that zone. The first 2 million characters (includes white space characters) per month for the first 12 months are free. From there, the price is $15USD for every 1 million characters per month.

I sampled three different types of text and translated them into Mexican Spanish using the real-time translate feature. My Spanish is limited to haggling at the beach, finding a bathroom, and order a soda. So how did I test if the translation was any good? I reversed the test to see if the Spanish translated to at least 98% of the English text – which it did. I also feel that the spirit of the text is 100%.

If you’re a native Spanish speaker, I would love to hear what you have to say about the translation!