All posts by Microsoft Translator

Customize Microsoft Translator’s Neural Machine Translation to translate just the way you want: Custom Translator now in General Availability

Custom Translator, now in general availability, significantly improves the quality of your translations by letting you build your own customized neural translation models tuned with your own pre-translated content.​ Using Custom Translator, you can translate your product names and industry jargon just the way you want.

With Custom Translator, an extension of the Microsoft Translator Text API, part of the Cognitive Services suite of products on Azure, you can build neural translation models that understand the terminology used in your own business and industry. The customized translation model will then seamlessly integrate into existing applications, workflows, and websites.

Custom Translator can be used with Microsoft Translator’s advanced neural machine translation when translating text using the Microsoft Translator Text API and speech translation using the Azure Cognitive Services Speech Service.

Preview customers of Custom Translator have already noted its improvements on translation quality and its usefulness regardless of the amount of pre-translated, bilingual content available.

Alex Yanishevsky, Senior Manager for machine translation at Welocalize, a leading language service provider, remarked, “Using Custom Translator, we’ve seen very good quality in comparison to other engines. It is very flexible. You can make engines just based on dictionaries if you don’t have enough data, and if you do have enough data you can make an engine based on data plus dictionaries. From the standpoint of customization, having that flexibility is really important.”

How it works

Custom Translator is easy to use and does not require a developer once the call to the Translator service has been properly set up in your app’s code. Custom Translator features a simple and intuitive web app that guides you through the 4-step process of customizing a model:

  1. Upload your data
  2. Train a model
  3. Test the model
  4. Deploy the new customized model to be used in your app

View the process in the image below.

For advanced use, there is also the Custom Translator API (preview) to automate the customization into your workflows.

Building and using custom NMT with Translator is quick, easy, and cost effective. By optimizing how training is performed, and how the Translator runtime incorporates the custom training, our team was able to provide a solution for customizing the Translator NMT models with a training cost that is less than 1% of the cost of training a new neural translation model from scratch. This, in turn, enables Microsoft to provide a cost-effective and simple pricing model to our users.

General availability pricing will go into effect on February 1st, 2019.

Get started now

  1. Ensure you have a Translator Text API key
    If you don’t have a key already, learn how to sign up.
  2. Log into the Custom Translator portal
    You can use your Microsoft account or corporate email to sign into the portal.
  3. Watch the how-to video and read the documentation.
  4. Questions?
    Ask them on Stack Overflow. We monitor these daily!

Go to Original Article
Author: Steve Clarke

Microsoft Translator launches Levantine Arabic as a new speech translation language

Microsoft Translator has released Levantine, an Arabic dialect spoken in countries such as Lebanon, Jordan, and Syria, as its latest AI-powered speech translation language. It will help businesses, educators, travelers, and non-profits communicate across the language barrier with Levantine speakers during meetings, presentations, and Skype calls.

credit: Photo of Beit ed-Dine in Lebanon by Oida666 from Wikimedia Commons

Levantine, our 11th speech language, is a spoken dialect of Arabic which has over 32 million native speakers.  Since it’s a spoken language that is rarely written, it lacks the large amount of parallel data required to train a usable machine translation system. As with any AI system, without the appropriate amount of data to train the neural machine translation model, the system won’t be able to produce translations that are good enough for real-life use.

However, our researchers developed a novel approach which utilizes monolingual data to train a system for any spoken dialect. This allowed the team to build a working Levantine to English translation system despite this lack of parallel data.

We adapted a system trained on standard Arabic-to-English translation to be used on a spoken Arabic dialect (Levantine) using only monolingual data of the spoken dialect. We developed an approach to generate synthetic parallel data from monolingual data.” – Hany-Hassan Awadalla, Principal Research Scientist 

Levantine is now available as a supported speech translation language through the Translator apps, Presentation Translator for PowerPoint, the Skype Translator feature in Skype for Windows 10, and the unified Speech translation service, an Azure Cognitive Service. With this service, developers can also customize speech transcriptions, translations, and text-to-speech, before integrating them into their apps, workflows, and websites.

Using the Translator app’s live conversation feature, users can have live, real-time conversations with people who speak other languages, on their own device, in their chosen language.

Let’s say you’re a Lebanese business person travelling to Italy and want to have a conversation with an Italian partner. You can speak Levantine into your phone or PC, and the Levantine audio will be translated into Italian text and speech on your partner’s phone or PC. This also works in reverse: the Italian speaker can speak into their device and have real-time multilingual conversations, and the listener receives the response in Arabic. This scenario is not limited to two devices or two languages. It can support up to 100 devices, across 11 speech translation languages, and over 60 text translation languages. To learn more about the Translator live feature go to or watch this how-to video.

Levantine speakers can also have translated, bilingual conversations using only one device by tapping the microphone icon and using the split-screen conversation feature in the app.  Simply select your speech languages, German and Levantine for instance, and use the app’s microphone button to speak in your chosen language. Translated text appears on the split-screen in each language.

Download the Microsoft Translator app.

Presentation Translator allows users to offer live, subtitled presentations straight from PowerPoint. As you speak, the add-in powered by the Microsoft Translator live feature, allows you to display subtitles directly on your PowerPoint presentation in any one of more than 60 supported text languages. This feature can also be used for audiences who are deaf or hard of hearing.

Additionally, up to 100 audience members in the room can follow along with the presentation in their own language, including the speaker’s language, on their phone, tablet or computer. This can also be used with the presenter’s language to support accessibility scenarios.

For example, if you’re presenting to a Levantine speaking audience and speak Spanish, you can choose Spanish as your speech translation language, and Arabic as the subtitle language. As you speak Spanish, your words will get translated to Arabic subtitling in real-time on the screen.

Levantine speakers can now also join and use their phone to ask questions, in Levantine, once the presenter unmutes the audience. This feature is useful for Q&A sessions after a presentation.

If there are audience members who speak other languages, they can follow along with the presentation in their chosen language in the Translator app or at

Levantine is also available for developers through the Azure Cognitive Services Speech service.  In addition to using the default speech translation models from Levantine, developers can also customize speech transcriptions and translation models using the Custom Speech ( and Custom Translator ( services.

Developers can then easily integrate speech translation into their apps using the new speech SDK available in several popular programming languages.

To learn more about Microsoft Translator for business, visit the Microsoft Translator site.

Microsoft Translator accelerates use of Neural Networks across its offerings

Exactly one year ago, Microsoft Translator made a new AI-powered technology available to everyone: neural machine translation (NMT) . Since then, the team has been working hard to expand the use of NMT to improve translations in the Microsoft Translator Text and Speech APIs, as well as in all of Microsoft Translator’s supported products.  

Today, the Microsoft Translator team is announcing several developments in NMT technology, making advanced AI translations more accessible, no matter how you use them:  

  • 10 new languages are now available for NMT in both the API and the apps 
  • All of the API traffic for Chinese and Hindi is now powered by NMT, developers do not need to do anything to use it 
  • New “hybrid” translation is available for API and apps users alike, bringing benefits of neural translations to languages not yet available on NMT 
  • Speech translation is now powered end to end with LSTM technology with the release of speech recognition LSTM systems in the speech API  
  • NMT is now also available as an on premises service 

NMT technology has transformed machine translation, providing major advances in translation quality over the existing industry-standard statistical machine translation (SMT) technology. 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 on the Microsoft Translator website, and watch the video below to see how NMT is integrated into Microsoft Translator’s speech translation.

[embedded content]

You can try out the new NMT powered translation today in Translator for Bing, the Microsoft Translator apps, Edge, the Translator PowerPoint add in, and of course, the Microsoft Translator Text and Speech APIs. 

Introducing 10 New NMT-Powered Languages 

Microsoft Translator adds 10 new languages to its list of NMT systems—now at 21 languages and growing! NMT now powers translations in the following languages (new languages in bold). 























100% of Chinese and Hindi translations moved to NMT  

Microsoft is making NTM more easily available to developers. Specifically, all Chinese simplified and Hindi translations to and from English are now using our new NMT systems by default.  This means that developers do not need to call the “generalnn” category in their apps to benefit from these new systems. For the other languages this category is still needed, but additional languages will be switched to 100% NMT in the coming weeks and months.  

For developers that want to keep using SMT systems for the time being, for instance if they have built customs systems using the Hub, the category “SMT” can be used when calling the API. 


“Hybrid” Neural/Statistical translation 

With hybrid translation, when only one of the two languages you are translating to or from is NMT-powered, you will still see an increase in translation quality.  

Because of the scarcity of available training data between languages and the exponential complexity of building and running dedicated systems for each language pair, machine translation systems use English as a “pivot language” to translate from one language to the other. This means that each time you ask a machine translations system to translate from, say, Chinese to Spanish, the translation system usually translates Chinese to English, then translates the English to Spanish.  

With hybrid translations, when at least one of the two languages has an available NMT system, Microsoft Translator will automatically use NMT for that section of the translation. This portion of the translation will improve, making the entire translation better. View our article on the Microsoft support forum to learn more about hybrid translation. 


End to end LSTM neural network powered speech translation to dramatically improve speech translation quality 

Speech recognition is moving to advanced LSTM neural network architecture. Combined with an increase in available speech data, LSTM speech recognition improves quality (measured by the industry standard “word error rate”) up to 29%, depending on the language. This has a direct impact on the quality of the machine translation, since the more accurate the speech recognition is, the more accurate the resulting translation will be.  

Microsoft Translator’s NMT uses LSTM technology speech translation is therefore now using  LSTM technology from end to end. Try out end to end LSTM speech translation capabilities with Microsoft Translator live feature in the Translator PowerPoint add in, on and the Microsoft Translator apps, or in the Translator Speech API. 

On premises neural networks  

For organizations that require additional data security, neural network translation is now also available as an on premises service, using the organization’s own servers rather than the Microsoft Azure cloud service. Learn more about Microsoft Translator’s on premises offering. 

Learn more in the blogs from Microsoft Research and Cognitive Services.

Microsoft Translator Adds Tamil as a Supported Language

Microsoft Translator is excited to announce the launch of our latest text translation language: Tamil.

Tamil is spoken by over 70 million native speakers throughout the Indian subcontinent and around the globe. Whether you’re using the Microsoft Translator apps, add-ins, or API on Azure, the addition of Tamil helps break the language barrier of communication worldwide and within the Indian subcontinent itself.

Meenakshi Temple. Madurai, Tamil Nadu, India

Communicating To and From Tamil: Apps & Add-Ins

Learn how to start communicating to and from Tamil through our apps and add-ins below. Click here to view our complete list of over 60 text translation languages – including other Indian subcontinent languages such as Bengali, Hindi, and Urdu – and our ten speech translation languages.

Microsoft Translator App

Text translation. Communicate to and from Tamil into over 60 text translation languages.

Multi-person conversation translation using the live feature. While traveling abroad, you can have live, real-time conversations with people who speak other languages, on your own device, in your chosen language. Let’s say you speak one of these ten speech translation languages, such as Spanish, and want to have a conversation with a Tamil speaker. You can speak Spanish into your phone, and the Spanish audio will be translated into Tamil text on the other person’s device! This also works in reverse: the Tamil speaker can type* into their device and reply to the conversation, and the Spanish speaker receives the response in Spanish. This scenario is not limited to two devices or two languages of course. To learn more about the Translator live feature go to

Phrasebooks for travel, directions, greetings, and emergency situations. Use the built-in Phrasebook feature for commonly used phrases for travel bookings, directions, dining, and much more.

Pin and save your most frequent translations for later. When you pin a frequently translated phrase, it is saved to your Favorites section within the Phrasebook. When you “star” or pin a translation, it will also appear in your Favorites.

Translate websites on Safari using the Microsoft Translator browser extension for iOS. When you download the Translator app on iOS, you automatically have access to the Safari translation feature. Open the app, tap Settings, and choose your Safari Translation Language. After choosing your language, open any localized website in Safari, tap the Share button, and the Microsoft Translator icon to translate the web content.

Translate text directly in other apps using the contextual text translation extension. There’s no need to switch to the Translator app to translate text! If you have the Translator app downloaded onto your device, this feature will automatically translate text within other apps. Simply highlight the text, tap the Share button, and tap the Microsoft Translator icon to see the translation.
Download the app on Android, iOS, or Windows.

Presentation Translator for PowerPoint (Windows only)

PowerPoint users can now display live, translated subtitles in Tamil by speaking in one of the 10 supported speech translation languages. Presentation Translator for PowerPoint gives audience members the opportunity to follow along on their own device, in their chosen text language. Download and learn more.

Translator for Outlook add-in

Translate email messages to and from Tamil across devices using an or Office365 email address. Read our blog for a complete list of features, and download the add-in here.

Translator for Microsoft Edge

Translate web pages to Tamil from over 60 other languages using the Microsoft Edge web browser for Windows 10. Download the extension here.

Translator for Microsoft Word

Translate entire Word documents into over 60 text translation languages. Translated text opens in a new Word document – maintaining original formatting, including tracked changes and comments!

Tamil Translation for Businesses

The Microsoft Translator text API helps with native translation support for solutions businesses market to sectors such as manufacturing, retail, education, and government services.

Part of Azure’s family of Cognitive Services AI-powered APIs, businesses can easily integrate Translator in their business process such as web localization, internal communication, or customer support.

To learn more about Microsoft Translator for business, visit the Microsoft Translator API website.

*Tamil speech to text not yet available.

Microsoft and Huawei deliver Full Neural On-device Translations

Microsoft is delivering the world’s first fully neural on device translations in the Microsoft Translator app for Android, customized for the Huawei Mate 10 series. Microsoft achieved this breakthrough by partnering with Huawei to customize Microsoft’s new neural technology for Huawei’s new NPU (Neural Processing Unit) hardware. This results in dramatically better and faster offline translations as compared to existing offline packs.

The Microsoft Translator app with these capabilities comes pre-installed on Huawei Mate 10 devices allowing every Mate 10 user to have native access to online quality level translations even when they are not connected to the Internet.

Until now, due to the computational requirements of neural machine translation, it was not possible to do full Neural Machine Translation (NMT) on-device. Huawei’s Mate 10 is the first phone featuring a dedicated Neural Processing Unit (NPU) for accelerating AI computing tasks. By combining Microsoft latest advancement in NMT on-device technology, and Huawei Mate 10 NPU, both speed and quality of offline translations are dramatically boosted. This offers users a unique offline language experience not found on other devices: the power of neural network translation without an internet connection.

What does this mean for users of the Microsoft Translator app on the Huawei Mate 10?

Imagine you are traveling and need to use the Translator app to ask for directions or understand a restaurant menu. With spotty network coverage and high roaming costs, you may not have an internet connection. Now, before you travel, you can download neural offline packs and have state-of-the-art neural translation at your fingertips, wherever you go.

“Before, customers had to choose between the quality of online and the convenience of offline translations. Now, with our NPU and Microsoft Translator software, customers get the best of both.”
Baofeng(Felix) Zhang, VP of CBG software and Head of AI, Huawei

AI-Powered Offline Translations using Neural Networks: The Numbers


Whether it is for its mobile app or its Translator API, part of Microsoft Cognitive Services, Microsoft uses the industry standard BLEU score to measure its translation quality. BLEU rates translation quality by comparing how close machine translations are to human ones.
Using this methodology, internal tests have shown, depending on the language, up to a 23 percent better offline translation quality over competing best-in-class offline packs, and often a difference of less than 1 BLEU point between Microsoft state-of-the art online neural and offline neural translations.

This means that even when you’re not connected to the Internet, offline translation quality using the Translator app on Huawei’s Mate 10 is closer to human translation than offline translations have ever been.


Quality of translations is not the only benefit of this new technology. Our tests also showed that translation of a full-page text picture was not only more accurate, but also up to three times faster than with the previous versions of Microsoft Translator on other high-end devices.


This new technology also saves space: The language packs for Huawei Mate 10 are 50 percent smaller than current Translator packs, leaving more space for your travel memories.

A full-featured app for all your mobile translation needs

The app has all the features Microsoft Translator is known for, including 60+ text translation languages and the Translator live feature that allows up to 100 people to have real-time conversations from their own device, including joining a translated presentation using the Presentation Translator PowerPoint add-in, and more. The full feature list can be found here.

The new Mate 10 exclusive neural translation packs are available for the following languages: Arabic, Chinese, English, French, German, Italian, Japanese, Korean, Portuguese, Russian, Spanish, and Thai.

In addition to these purely neural network powered languages, additional language packs are available for the remaining languages.

Read more about the Huawei Mate 10 release here: