← All events

Accelerate Global Course Delivery with AI Translation

August 20, 5:39 PM
YouTube video player

 Breaking Through Barriers and Increasing Learning Accessibility

Topics covered in this webinar

➡️  Why language matters
➡️  Breaking through the globalization bottlenecks
➡️  Customer Story - how Intradiem took their training global in under 6 weeks with Smartcat
➡️  Smartcat AI demo

Webinar Q&A

Q: What happened to translations that have branded terms that need to remain in English?

A: That's not a problem for Smartcat. You can create glossaries that can be associated with any file or projects, and any terms that need to remain in English will be highlighted inside the editor. These highlights can then be used as a cute for the translator working inside the editor -- they'll be able to see the terms that need to remain in English and ensure they don't translate. if they inadvertently translate these terms it'll generate a QA error which alerts the project manager that the the document needs to be reviewed and and corrected.

Q: Can we teach Smartcat some nontypical translations that we use in our organization? For example, there are words that we always use in English, even if we are writing in Spanish.

A: This is similar to the question above about branded terms. You can add these words or terms to your own glossaries of terms, and Smartcat will identify and use those. This applies to both manual translation and also AI translation. We didn't go through this today but our AI translation engines can be trained to use your own terminology when generating the transition.

Q: Can you use or connect to an internal translation engine by policy? We're not allowed to use DeepL or Google Translate.

A: Yes, you can bring your own transition engines and limit the use of other engines. You can also use our API to integrate with whatever translation engine you want.

Q: Articulate Rise has a lot of character limits for different interactions, How do you recommend handling that? What is the best way to identify those sections?

A: Inside the Smartcat Editor we can show you the number of characters that you have in your document or segment. So you can set limits and provide instructions to your freelancers or translators. So you could limit the translation to be the same length as the source language, or limit the allowed expansion by a certain percentage -- 10%, 15%, whatever. And again, this is also checked by Smartcat's AI-powered AW process, so if the document or segment exceeds a certain length, it will generate an error, and the translator will need to correct the segment before it goes through to final confirmation. So there are many different ways to to handle this, but it's easy to fix in Smartcat.

Q: If we use the AI Assistant to generate content, does that become available to others? i.e. do we have intellectual property considerations to think about?

A: When you create something original, or create a translation of an existing piece, it is all gated inside your own Smartcat account. There is no there's no sharing of data between your account and anyone else's account, and that's also true for what I was showing you the AI Assistant. The only people that have access to your data are the people that you explicitly grant access to. Not even Smartcat people can access your data.

Q: Could you go into more detail on how Translation Memory works? How the data is stored, how it can be customized and used to give the best possible translation?

A: Every time you translate a sentence inside the Smartcat Editor, that sentence gets stored into the translation memory database. When you translate another document, Smartcat will first look for similar or identical strings or sentences in the database. If we find a sentence in the database where all the words match, we'll call that 100% match and we will bring this sentence into your new document. We also look for what we call fuzzy matches -- for example, if there were minor variations between the sentences we could bring that in but mark it with a lower certainty or match in the editor. We also we also store context, meaning the sentence before and sentence after the one we're matching for. So if you build something from the transition memory database, there's greater certainty that the translated content matches what you require.

If you want to learn more, contact us and we can give you a more in-depth demo and show you how the system works.

Q: How are images translated? Can you work with layered Photoshop files?

A: Smartcat does not directly support translation of Photoshop files, it's not an easy format to work with honestly. If you upload a jpeg or other bitmap image file into Smartcat we can extract the text using optical character recognition (OCR). You would then need to manually update you Photoshop files with the translated text.
We work with a lot of other image and design file formats however. You can bring you Adobe InDesign files directly into Smartcat and we can edit those and retain all the formatting.

We also have a direct plug-in with Figma. So you can send content directly to Smartcat from within Figma, and immediately get the translated copy back and test it for fit, line breaks and so on.

Q: Is the Articulate support (and any of the other plugins mentioned) included in the core product, or are these extra pieces that need to be acquired at additional cost?

A: These are included in all product levels, so you can translate Articulate files with event the basic system.

Q: Why do we have to export Articulate files to Smartcat? And is there anything specific that I need to do to prepare those articulate files for Smartcat?

A: There's no special preparation needed. You just go into your Articulate course, export the XLIFF file, and translate it in Smartcat. Articulate doesn't expose their API, which is what's needed in order to build a direct integration. Everyone has the same challenge with Articulate, this isn't unique to Smartcat. If we ever get API access we will build that direct integration! In the meantime we've made it as seamless as possible for Articulate users to get high quality results. As you saw we use a special XLIFF variant for Articulate, so there's no errors when you bring your files back into Articulate.

Q: What video file formats do you support?

A: I think MP4 is the only one that we support. I'll validate if we support .MOV. We also support MP3 for audio.

Q: How many languages does Smartcat support?

A: 362, and counting

Q: Do you support Canva files?

A: So Canva is also one of those applications that doesn't make it easy to translate! They have an API available, but no API that can be used for integration. This is something we ran into with Cynthia and the Intradiem project. Best suggestion is to export the Canva file as a PDF with editable text, which can then be brought into Smartcat for translation. Then you can put it back in Canva, or just use the PDF as-is if the formatting is okay. If it is very graphically intensive you could export as an image instead, and then we can extract the text using OCR.

Learn how your organization can scale resources dynamically

With the power of AI sourcing and AI human workflows