Regular host Louise Law chats with Mandy Morgan, Strategic Account Manager at Welocalize, to explore how AI-powered tools can revolutionize multimedia localization for global communications and e-learning.

Mandy shares her expertise and dives into a case study where AI solutions potentially reduced costs by an astonishing 86% for a leading tech company.

Discover how condensed multilingual recap videos, AI avatars and captions, and innovative AI-enabled workflows are transforming the accessibility and efficiency of training and corporate content. This episode is a must-listen if you’re working with global multimedia and e-learning content for multilingual audiences.

---

TRANSCRIPT 

Louise Law: Hi everyone. Welcome to the Welocalize podcast, where we talk about all things relating to global content, localization, translation, and the emerging AI technologies that are transforming how we create multilingual content. Today on the podcast, we’ll dive into the world of AI-powered multimedia localization with Mandy Morgan from Welocalize, who has worked in the language industry for many years. Mandy works with a lot of brands, helping them develop their localization translation program. As well as sharing her thoughts on the AI tools that are reshaping how we localize multimedia, we’ll unpack a recent real-life case study that Mandy recently worked on where the use of AI cut costs and a lot of communications to deliver corporate videos to a global and culturally diverse workforce. So, first and foremost, Mandy, welcome to the podcast.   

Mandy Morgan: Thank you, Louise. It’s very nice to be here.   

Louise Law: Yeah, it’s lovely to have you. Do you think you could tell the listeners a little bit about yourself and your role at Welocalize?   

Mandy Morgan: Absolutely. I’ve been in the localization industry for over 18 years. 9 1/2 of those have been with Welocalize. Over those years, I’ve worn many different hats. I spent the majority of my career in operations, and about two years ago, I moved over to account management. I really enjoyed the move, it’s been great. I enjoy working really closely with the clients. There is a lot of creative solutioning that happens and understanding their pain points, but most of all, I really enjoy partnering with them.   

Louise Law: We’re here to talk about multimedia localization. Mandy, can you briefly explain what that is and why it’s so important for global organizations?  

Mandy Morgan: Of course. Yeah, multimedia localization is really adapting various types of multimedia content, such as videos or audio graphics, into multilingual formats. So, this is becoming more and more evident as I work with my clients that video localization is a strategic investment for their organizations. Some of their goals include effective communication, engaging with their global audience, and just growing in diverse markets. Video does play a pivotal role in the language and cultural gaps as well. But their goal is to make it more impactful and relevant for their audiences around the world.   

Louise Law: What are some of the challenges faced by brands today in getting multimedia content into multiple languages?   

Mandy Morgan: What I hear time and time again is time and cost, and I think we hear that across the industry, so it’s not specific to multimedia. But multimedia is very time-consuming. Clients often work with really tight deadlines and to produce these, it takes a very meticulous planning and coordinating process, so project managers are key in this. In addition, it’s expensive because there are so many different components. When we talk about voiceovers, subtitles, on-screen text, and graphics, you multiply that by the many languages and regions that they’re targeting. We also find that there are a lot of specialized tools and software, and then adding a human component like voiceover and the retakes just adds extra cost.   

Louise Law: I know you work with many well-known global brands, and you’ve been working with several organizations to integrate AI into their programs and, in particular, into how they approach global multimedia. I know you have a particular technology client who had, I think, over 60, 70, or 80 videos that they needed translating into seven languages. This is just a perfect example of how complete, full-scale, no-expense localization for such a high volume of videos could really blow the budget. So, how did you creatively solve this and integrate AI to help them?   

Mandy Morgan: No, it’s funny; when they came to us with this project, we didn’t think about AI at all. Then I started thinking, OK, 60 videos is roughly an hour apiece. Who is going to watch all these videos? So, their goal was to go into seven languages. So, I took the script and ran it through – we have our own version of ChatGPT here at Welocalize. So, I ran the scripts through that and asked them to summarize and make a three to five-minute script. From there, we ran it through a machine translation. We did some fine-tuning and prompting to make sure that we were capturing the right tone and style. And then we ran it through synthetic voiceover, so taking out another human component. There, we decided to sync avatars. So, these were live videos, and now we have generated an avatar that speaks the summarized script. We went from about $1,000,000 for this project down to a couple of thousand. It was really fun. The goal of it was to produce these short, fun, digestible videos at a fraction of the cost.   

Louise Law: That’s a great example of how AI tooling can be used to reduce cost, speed up the workflow, and make it a possible option for organizations rather than not being able to do it at all. I know you mentioned before that the initial cost projection using traditional methods could have been up to $1,000,000. How common is this for large-scale video localization when those costs are so high?   

Mandy Morgan: It’s actually really common when you think about the complexity of multimedia and e-learning. It adds up really quickly, and the files are really nuanced; they’re very particular. When you take into consideration animation, on-screen text, and then, as I mentioned earlier, that traditional approach of the human voice over. And any retakes, any errors they make, we have to get them back in the studio and rerecord. I would say, yeah, it’s very common; those prices add up quickly and then you multiply it up into multiple languages.   

Louise Law: And it sounded like you implemented an AI element at various touch points in this challenge. Could you tell us a little bit more about the role that AI played? How did these solutions work?   

Mandy Morgan: Yeah, we actually integrated AI into almost every aspect. There is a benefit of having a human in the loop, so having a human review and make sure that the content is appropriate. As I mentioned, we ran it through the ChatGPT to get a summarized script and that was in English. We then had a linguist review it. We did some prompting to get that tone and style correct, but once we were satisfied with the English script, we ran it through the machine translation. We used LLMs to post edit and then we did the synthetic voice and the avatars. Once it was all compiled together, we did have a human review just to make sure that it was appropriate.   

Louise Law: It sounds like kind of a nice balance, and your projected results were, I think, you were telling me it was up to 86% cost savings, right?   

Mandy Morgan: Yeah, I think originally, each video was about 9 or 10,000, and we got it to just around $1000 per video. So, still human, there’s still some cost to AI, so it’s not free, but it is a significant cost reduction.   

Louise Law: And obviously, how would you say this approach improves the user and the learner experience? Videos they use to help educate and for corporate audiences and clients to learn. How would you say this improved the overall experience using this approach?   

Mandy Morgan: Well, it gave them the ability to digest a lot more videos. Everyone is really pressed for time, so who has the time to watch an hour-long video? I would also like to say that, obviously, we wouldn’t do this for like a keynote or CEO speech, but just to get an understanding of what these videos are and what the conference was about, I think it was really successful.   

Louise Law: You just get the gist of it, don’t you?   

Mandy Morgan: Exactly, yeah.   

Louise Law: You mentioned using AI avatars and unedited captions and subtitles. These can potentially feel unfamiliar and maybe not be personal enough; you know, people might spot that it doesn’t have full production value. How did you ensure credibility and cultural appropriateness for the finished product?   

Mandy Morgan: Surprisingly, they’re really realistic. But we did add some disclaimers and explanations stating that they were avatars. We did have a human go through and make the adjustments to make sure that that tone, especially the style of the client and who their target audience was, making sure that was all captured. And then we did have a little blurb at the bottom saying that these were avatars and synthetic voices and were not the real people that we are trying to represent from the company, so we tried to make it clear, but it would also really fun use to bring AI into their workflows.   

Louise Law: It’s a great use case for them as well, kind of demonstrating how they’re using AI. Streamline and do things better and cheaper, really. Are there any specific areas or industries where we would work on multimedia localization that are benefiting the most from AI in this area?   

Mandy Morgan: I think anywhere that you know, it drives consistency, and it drives efficiency. So anywhere that you can do that would be beneficial. When I try to think about like specific areas, the e-learning and educational content is good. Social media, there’s so much content that’s user-generated, so platforms, YouTube, Instagram, TikTok, those types of things, AI could play a huge role in getting large quantities of content translated quickly, and then as we’re talking about today, like corporate training and videos, live events, those work really well with AI. It depends on who your audience is, but I think it’s a really good place to start.   

Louise Law: So, what’s next in this space? How can organizations get started?   

Mandy Morgan: Well, as we all know, AI is advancing very quickly. I keep hearing more and more customers having mandates to introduce AI into their workflows, so I would recommend doing some pilots, reaching out to your project managers or account managers, talk about different potential projects that you could have coming and talk about how you could introduce it. You can reach out to myself or Louise. Happy to help you, but I think just exploring and embracing. It’s new for a lot of people and it’s kind of scary until you dive in. So, I just say embrace it and have fun with it.   

Louise Law: I think you just have to experiment and just get started and take it from there, don’t you?   

Mandy Morgan: Exactly. And I think going into you know the coming up years we’re going to see a lot more mandates of trying to introduce it, reduce costs, produce higher volumes quicker. AI is going to help aid in a lot of those discussions.   

Louise Law: And there’s some great AI tooling out there isn’t there. So, I think it’s kind of key to collaborate and keeping an eye on what the tools are, how they can be integrated into the overall solution.   

Mandy Morgan: Yeah, I think we’re experimenting almost on a daily basis, here at Welocalize.   

Louise Law: For sure, yeah. Mandy, are there any specific areas of multimedia localization that are really benefiting a lot from AI?   

Mandy Morgan: I think anywhere you can incorporate AI will drive efficiencies, reduce cost and then drive consistency across your localization content. A few areas that come to mind are automated transcription or translation, so AI can quickly generate transcripts versus having to listen and type back in the old days. Quality Assurance. So, AI drives quality, it detects errors, inconsistencies, cultural issues, and localization content is also really important. We’re seeing a lot of improvements in synthetic voice and dubbing, so AI can produce synthetic voice and fine-tune it for accent, tone, and emotion, which is pretty incredible. When I think about content areas, I would say e-learning and educational content are a great place to start. When I think about content areas, I would say e-learning and educational content are the vast amount. So those would be a great place to use AI to translate high volumes. And like corporate training and communication. I think all corporations that are global are trying to reach their audiences and their employees across the globe. So, using AI to aid that will help keep costs down, but allow you to reach your entire global staff.   

Louise Law: Mandy, thank you so much for coming in and being a guest on the podcast. Really appreciate your time, really appreciate your expertise. So, I’d just like to say thank you for joining us today.   

Mandy Morgan: Perfect, Louise. Thanks for having me.