Frequent questions about LanguageWire MT API

General

What is an API response?

An API response is data sent back by a server in response to a call or request you made.

Are there different API responses?

Yes, you can get a success or error responses.

What are the available API environments?

We have only a production available for our users.

Access

How can I start using the MT API endpoints?

Our quick introduction guide will go over how to get started with text translations.

How can I get the access credentials needed for the MT API?

If you want to try out the API, you can do so with trial credentials. Alternatively, you can generate a set of production credentials. 👉 Learn how to get your credentials here.

Can I have access to all MT API endpoints?

Yes, you can use all endpoints as long as you have an access key. 👉 Learn how to get your credentials here.

When does a token expire?

Tokens are temporary and expire after 20 minutes. You should generate another token after that time.

I already have an API key. Can I continue using it?

We are phasing out the support for API keys. All new integrations should be created using OAuth client credentials. We will continue supporting API keys for existing integrations until they have migrated to the new credentials.

Dashboard applications

When should I delete my app?

You should delete you application from our Dashboard in these cases:

• When the integration is no longer needed.

Documents

Are there any restrictions in documents?

Yes, there are. Refer to the documents management guide for more details.

Languages

Which languages are available for the MT API?

You can see the full list of available language pairs by using the GET /v2/languages/text-translation and GET /v2/languages/document-translation endpoints.

Which languages codes does the MT API use?

The languages are triggered using the codes defined by the BCP 47 format.

👉 Refer to the languages standards and formats section for more details.

MT Engine

Does the MT quality improve over the time?

Yes! Our generic MT engines are trained only on high-quality data, i.e. human translated/verified content. The re-training is happening on a regular basis to ensure continuous improvement. The more Translation Memory data a user produces with us, and the more similar content is submitted from that user, the better the MT engine is at handling the content, i.e the better it gets the tone of voice and terminology.

How often are the MT engines re-trained?

We are re-training our MT engines on a regular basis, usually every month. The period can also depend on how much new data we were able to collect.

Something missing? If you have any other related question that is not listed here, and would like us to respond, share it with us.