The new Microsoft Translator text API version 3 is now available for developers. It natively uses neural machine translation, offers a modern API implementation, and makes it easier than ever to add advanced translation to your sites, apps, and workflows.

 

What’s New

The latest V3 release is the first end-to-end update of the API itself, offering several new translation features and capabilities:

• Natively neural: Modern neural machine translation (NMT) is the default for all supported languages. NMT better captures the context of full sentences before translating them, providing much higher quality translation and more human-sounding output. Learn more about how NMT works.
• GDPR compliant as a processor
• Support for the new Translator custom feature which enables users to build and deploy customized neural translation systems.
• Translate into multiple languages at once– V3 allows the application to translate into multiple languages with a single GET request. The previous version required separate API calls to translate the same text in multiple languages.
• Transliteration– Convert words and sentences from one script into another script. For instance, if your users translate from English to Arabic, they may not be able to read the Arabic script. The new transliteration feature allows you to offer the Arabic words written in Latin (i.e. English) letters.
• Bilingual dictionary– The new bilingual dictionary feature displays alternative translations from or to English, examples of words in context to help you choose the perfect translation, and even examples of translated sentences using this word. Already available on Translator for Bing, and in the Translator apps, this feature is now accessible to any developer.
• JSON formatting– All data sent to and received from the service is now in JSON format.

 

PubNub, a company that offers real-time infrastructure-as-a-service, has partnered with Microsoft Translator and has integrated Microsoft Translator’s latest version in its offering. Now, developers can leverage Translator’s newest features and capabilities through ChatEngine, an open and extensible chat SDK and APIs for building powerful 1:1 group chats quickly.

We’re excited to partner with Microsoft to bring AI-powered language translation to the world’s largest chat deployments. Through an integration with PubNub ChatEngine, it is now possible for any size enterprise to easily enhance their chat apps with powerful trans-lingual capabilities, while meeting the most demanding scaling, security and reliability requirements.  – Todd Greene, Co-Founder & CEO

How to get started

The new version is available through a regular Translator text API Azure subscription. To access the new feature, you can use your existing key and only change the appropriate parameters: end-point and API calls. Both the subscription and pricing tiers remain the same.

For existing V2 API users:

1. Review the V3 documentation at www.aka.ms/Translatorv3Docs
2. Learn how to modify your V2 code to use the V3 API at www.aka.ms/TranslatorV3migration
3. Use existing sample apps from the Translator Github repo to get you started

For new users: Follow the steps on our Get Started Page. It will also provide you with the complete documentation, sample code, and support contact. https://www.microsoft.com/en-us/translator/trial.aspx

Following the general availability of Version 3, the existing Version 2 will be deprecated starting May 1st. V2 will remain supported until 04/30/2019.

Translator V3 text API comes with other Translator product releases for Microsoft Translator:

• The custom feature allows for not only customized text translation, but speech translation as well. With as few as 2,000 parallel sentences, i.e. human generated translations, you can begin customizing your translations.
• A preview of the new unified Cognitive Services Speech which combines the capabilities of the existing Translator Speech API, Bing Speech API, and Custom Speech Service in a unified and fully customizable set of services: speech to text, translation, and text-to-speech

Learn more

Customized neural translation
How neural machine translation works