You have a document, a file, or a piece of text that you need to quickly translate from one language to another. Most probably, you will go to google translate or any other free machine translation engine, drag and drop the file or paste the text, and after a few seconds, it is translated. Quality isn’t of importance, you just want to understand what the text says.
A fast, convenient, and free translation service. But is it really free? Well if you were to read the Terms & Conditions of the service, you might think otherwise. So we are here to help! We created a short overview of a few popular machine translation engines: Google Translate, Microsoft Translator, IBM Watson, and Yandex. But first, we will give you some background.
An intro to Neural Machine Translation
As many of you may know, Neural Machine Translation (NMT) technology has transformed machine translation. NMT has been providing significant advances in translation quality over the existing industry-standard, statistical machine translation (SMT) technology. It captures better the context of full sentences before translating them, providing much higher quality translation and more human-sounding output.
Furthermore, by using specialized NMT on-line services, you can:
- Translate into multiple languages with a single request. The feature automatically detects the source language of the content.
- Convert words and sentences from one script into another script. For instance, you can present Chinese characters in the Latin alphabet (PinYin) so they can be read by non-Chinese audiences.
- Display alternative translations from or to English, examples of words in context to help you choose the perfect translation, and examples of translated sentences using this word.
The easy access to free translation services may provoke you to investigate a bit more what each translation engine is “hiding” in their Terms and conditions section.
Please keep in mind, however, that we are referring only to the free versions from these providers. They, however, offer paid versions as well (at Text United we use only the paid versions). Paid versions include much broader sets of functionalities as well as security and confidentiality options. Let’s begin.
Microsoft claims that it meets strict security requirements. Customer information that is submitted for translation by Microsoft Translator is not written to persistent storage and there will be no record of the submitted text in any Microsoft data center. Microsoft Translator has also received compliance certifications such as HIPAA, ISO, SOC, and FedRAMP.
Since Microsoft Translator has two different translation approaches, we would like to explain both. The first, Standard Translation is an easy one-step process, which you provide an input sentence, and Azure’s pre-trained models translate them into a target language. The second approach, Custom Translation is a two-step process, which first you train your models using training data you upload to the translation service. Once your model is trained, you can perform the translation as you would with Standard Translation.
When it comes to the Custom Translation, there are three billed activities that you can perform:
- Model hosting: Hosting a model means that it is available to use for Custom Translation. You are charged a flat fee for every model that is hosted during a billing period. This is NOT pro-rated if the model is hosted for less than the full month.
- Training: Every time you train a model, you are charged a fee for every character in the training data. Characters in both the source and target languages of the training set will be charged. However, there is a cap to how much you can be charged for any given training run, no matter how many characters are in your training data. This cap applies to each training run. For example, you will be charged if you will re-run the same set.
- Translation: You are charged for every character of text translated by your Custom Translation model.
With Microsoft Translator, free billing and subscription management support are included. When it comes to service credits, they are your sole and exclusive remedy for any performance or availability issues for any service under the agreement and SLA. You may not unilaterally offset your applicable monthly service fees for any performance or availability issues.
Service credits only apply to fees paid for a particular service, which did not meet the adequate level. The service
credits awarded in any billing month for a particular service will not exceed your monthly service fees for that service in the billing month.
If you purchased Services as part of a suite or other single offer, the Applicable Monthly Service Fees and service credit for each Service, will be pro-rated. However, if you purchased a service from a reseller, you would receive a service credit directly from your reseller, and the reseller will receive a service credit directly from Microsoft Translator. The service credit will be based on the estimated retail price for the applicable Service, as determined by Microsoft at their reasonable discretion.
Of course, there is one important element that we should underline for people who are not willing to move forward to the paid plans. The free option offers you 2 million characters of any combination of standard translation and custom training per month.
When it comes to IBM, you can use Watson on the IBM Cloud and can combine your data sets with other IBM-owned, licensed, or public data sets to yield even more valuable insights. IBM makes clear when and for what purposes your data is being applied in the solutions they develop and deploy. This may include the major sources of aggregated data and expertise that inform the insights of AI solutions IBM develops. Third-party or licensed data will be identified. When you use Watson services to enhance your business processes, the unique insights you uncover are your competitive advantage.
Also, IBM agreements are transparent. They claim that they will not use your data unless you consent to such use. Only then, IBM will limit the usage of your data to specific use cases, clearly described in the agreement. Furthermore, IBM will not share unique insights derived from your data without your consent. IBM will return or destroy client content at the request of a client or the end of the cloud service term.
Before we elaborate on Yandex’s Terms and Conditions, here’s a little interesting fact: Yandex was initially introduced to translate Wikipedia pages from English to Russian. It is now extended to all languages.
Here are their T&Cs:
No personal information is sent to Yandex.
The TMT system will be accessed via a publicly available API using a key. Article content (freely licensed) is sent to Yandex servers from Wikimedia Foundation servers. No direct communication is happening between the user and external services, and no personal information (IP, username) is sent to Yandex servers. The client contacting Yandex servers is open source, and you can check it here. No part of Yandex’s service or code will be part of Wikimedia infrastructure or Content Translation codebase.
Information is returned from Yandex under a free license.
When Yandex service is used, they provide a translated version of Wikipedia content under a free license. Users can modify it and publish it as part of Wikipedia without conflicts with existing policies. The resulting content translated by Yandex and the user modifications will be available under the same license that is used for the rest of the articles in Wikipedia.
Benefits of the wider open-source translation community.
Translations obtained from Yandex and user modifications will be publicly available. The post-edited translations are especially interesting for the translation research community. They can use this resource to create new translation services to support languages for which open-source machine translation is not available yet. Also, this will help developers create and improve machine translation systems.
Users can disable it.
Automatic translation is an optional tool in Content Translation. Users have an option to disable it if they don’t find it useful. Although many users from the Russian community requested this service, each user eventually decides whether they would like to use it or not.
It is the most common and used free machine translation engine out there. The paid version of Google Translate services allows you to upload, submit, store, send, or receive content. You retain ownership of any intellectual property rights that you hold in that content. In short, what belongs to you stays yours.
On the other hand, when you upload, submit, store, send, or receive content to or through free Google’s services, you give Google (and those who they work with) a worldwide license to use, communicate, publish and distribute such content. The rights you grant in the license are for the limited purpose of operating, promoting, and improving Google Services, and to develop new ones.
You can also access and remove content that was provided to Google. Also, some terms or settings narrow the scope of their use of the content submitted in those services. Make sure you have the necessary rights to grant Google this license for any content that you submit to their services.
When it comes to the terms of billing the services, you are charged for Cloud Translation based on monthly usage. Google states usage fees for Cloud Translation – Basic and Cloud Translation – Advanced. Translation usage is calculated in millions of characters, where 1 million = 106 characters. To calculate usage, Google counts usage on a per-character basis, even if a character is multiple bytes. Each character corresponds to a code point.
You are charged for all characters that you include in a Cloud Translation request, even untranslated characters. This includes, for example, whitespace characters.
If you translate こんにちは to English, it counts as 12 characters for billing. Google also charges for empty queries. If you request without any content, Google charges one character for the request.
In summary, all the Free Machine Translation engines on the market offer free translation but with specific terms and conditions (or limitations). We gave you an overview of the highlights, however, make sure that you completely understand the engine’s terms before you go ahead and upload a file, or paste a text for translation.