New SAG-AFTRA Game Localization Contract Restricts AI Usage in Dubbing

On November 14, 2024, the Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA) announced an updated version of a previous agreement that covers the localization of video game projects produced in a non-English language.

With approximately 160,000 members, SAG-AFTRA calls itself the “world’s largest union representing performers and broadcasters.” The union also represents voiceover artists, including those who provide dubbing. 

As comics and gaming website Bleeding Cool reported, the new Independent Interactive Localization Agreement is essentially an updated version of the base terms from the union’s Tiered Budget Independent Interactive Media Agreement, plus AI protections.

The new agreement is signed on a project-by-project basis by employers whose project was originally scripted in a language other than English, and whose intellectual property owner is based outside of the United States.

“Many brilliant, beloved games come to market in the U.S. from other countries, projects which need highly skilled localizing performers,” Interactive Media Agreement Negotiating Committee Chair Sarah Elmaleh was quoted as saying in the press release. Elmaleh added that “[m]any such companies have already signed Interim Localization Agreements”. 

The contract was reportedly crafted “based on direct feedback from the community that does this work.”

However, the introduction of the Independent Interactive Localization Agreement does not interrupt an ongoing SAG-AFTRA strike.

The union called for the strike, effective July 26, 2024, in response to stalled negotiations, which began in October 2022. The use of AI still presents a major hurdle, as SAG-AFTRA explained:

“Although agreements have been reached on many issues important to SAG-AFTRA members, the employers refuse to plainly affirm, in clear and enforceable language, that they will protect all performers covered by this contract in their A.I. language.”

The strike applies to over 130 video game projects currently signed to the union’s interim and independent agreements; the goal is for those signatories to sign updated agreements. 

Human-Made Recordings and Digital Replicas

The Interactive Media Agreement, originally introduced in 2017, was extended until 2022. At that point, it was replaced by the Tiered-Budget Independent Interactive Media Agreement along with the Interim Interactive Media Agreement and, most recently, the Independent Interactive Localization Agreement.

Katie Sikkema, a union contracts consultant, explained on LinkedIn that the updated agreement “includes a mandatory buyout of all reuse and integration for an additional 50% of scale, which means the IP owner can use the (original, human-made) recordings for whatever they want with no further fees due (except merchandising/talking toys, which must be separately negotiated).”

The contract describes generative AI (GenAI) as a “subset of AI that learns patterns from data and then produces content based on those patterns.” 

Interestingly, GenAI is considered separate from “digital replicas” of performers, including voice actors, though the term is not defined. To that point, employers are required to provide the union advance notice if they intend to use GenAI to generate material other than digital replicas. 

Employers that want to create GenAI material using prompts including a performer’s name, or a unique character associated with that performer, must get consent from the performer and bargain for the use of GenAI material at a specific minimum rate.

SAG-AFTRA’s strikes have brought attention to the issues GenAI presents for the entertainment industry, an area ripe with opportunities for language AI startups specializing in dubbing. In March 2024, the union ratified the 2023 Television Animation Agreement, which includes “strong protections around the use of artificial intelligence” for voice acting and other performances. 

Google Says There’s a Better Way to Create High-Quality Training Data for AI Translation

In an October 14, 2024 paper, Google researchers highlighted the potential of AI translations refined by humans or human translations refined by large language models (LLMs) as alternatives to traditional human-only references.

Talking to Slator, Zhongtao Liu, a Software Engineer at Google, explained that their study addresses a growing challenge in the translation industry: scaling the collection of high-quality data needed for fine-tuning and testing machine translation (MT) systems. 

With translation demand expanding across multiple languages, domains, and use cases, traditional methods that rely solely on human translators have become increasingly expensive, time-consuming, and hard to scale.

To address this challenge, the researchers explored more efficient approaches to collect high-quality translation data. They compared 11 different approaches — including human-only, machine-only, and hybrid methods — to determine the most effective and cost-efficient one.

Human-only workflows involved either a single human translation step or included an additional one or two human review steps. Machine-only workflows ranged from single-step AI translations using top AI systems — MT systems or LLMs — to more complex workflows, where AI translations were refined by an LLM. Hybrid workflows combined human expertise and AI efficiency; in some cases, AI translations were refined by humans (i.e., post-editors), while in others, human translations were refined by LLMs.

They found that combining human expertise and AI efficiency can achieve translation quality comparable to, or even better than, traditional human-only workflows — all while significantly reducing costs. “Our findings demonstrate that human-machine collaboration can match or even exceed human-only translation quality while being more cost-efficient,” the researchers said.

The best combination of quality and cost appears to be human post-editing of AI translations. This approach delivered top-tier quality at only 60% of the cost of traditional human-only methods, while maintaining the same level of quality.

“This indicates that human-machine collaboration can be a faster, more cost-efficient alternative to traditional collection of translations from humans, optimizing both quality and resource allocation by leveraging the strengths of both humans and machines,” they noted.

The researchers emphasized that the quality improvements stem from the complementary strengths of human and AI collaboration, rather than from the superior capability of either the AI or the human (post-editor) alone, underscoring the importance of leveraging both human and AI strengths to achieve optimal translation quality.

They noted that LLMs were less effective than human post-editors at identifying and correcting errors in AI-generated translations. On the other hand, human reviewers tended to make fewer changes when reviewing human-generated translations, possibly overlooking certain errors. Interestingly, even additional rounds of human review did not substantially improve the quality. This observation supports the argument for human-machine collaboration, where each component helps address the other’s blind spots, according to the researchers.

“These findings highlight the complementary strengths of human and machine post-editing methods, indicating that a hybrid method is likely the most effective strategy,” they said.

Authors: Zhongtao Liu, Parker Riley, Daniel Deutsch, Alison Lui, Mengmeng Niu, Apu Shah, and Markus Freitag

Leading Localization from Asia with EC Innovations’ Sijie Wei

Sijie Wei, Co-CEO of language services, technology, and game localization specialist EC Innovations (ECI), joins SlatorPod to talk about his new role as co-CEO and leading one of Asia’s largest LSPs with revenues exceeding USD 50m in 2023.

Sijie describes how ECI's initial focus on assisting Western blue-chip companies to enter China evolved into helping Chinese digital titans like Tencent and ByteDance extend their product offerings internationally.

Sijie emphasized the difficulties facing the Asia-Pacific market, where localization maturity in several verticals still lags behind that of the US and Europe. But as sectors like e-commerce, artificial intelligence, and electric vehicles develop, he sees enormous development potential.

Sijie noted that Chinese businesses want to create their own language AI solutions, which makes the industry extremely competitive. This study helps ECI apply state-of-the-art AI developments to client projects. 

https://youtu.be/PrJkZEWWCN0

In game localization, Sijie continues to see substantial growth potential. He recognizes that localizing games from China to global markets and vice versa is equally challenging due to differences in player preferences, monetization methods, and cultural contexts.

Sijie shared his thoughts on the financial environment, pointing out that the uncertainty surrounding the adoption of AI in several industries, such as localization, is the reason why market values are currently lower.

Sijie addressed the balance between AI and human localization knowledge in her conclusion, stressing that although AI can increase efficiency, human quality is still preferable in many situations.

Smartcat's Series C Funding, YouTube Dubs Launch, Viva Translate Closes Down

Slator- Language Industry Intelligence

Florian and Esther discuss the language industry news of the week, where they give their impressions from SlatorCon Silicon Valley and touch upon the findings from the 2024 ALC Industry Survey.

In a significant funding update, Esther reports that Smartcat raised USD 43m in a Series C round, bringing their total funding to USD 70m. This funding will support product innovation in AI translation and multilingual content generation.

Florian talks about YouTube’s potential launch of AI dubbing, a feature in testing that aims to generate translated audio tracks for videos, significantly enhancing content accessibility and engagement.

In Esther’s M&A corner, Cloudbreak, now rebranded as Equiti, acquired its competitor Voyce and brought on a new private equity partner, Heritage Group. Meanwhile, EasyTranslate acquired World Translation, expanding its reach in the Nordic and DACH regions.

The duo bid farewell to publicly traded Keywords Studios, which is delisting after being acquired by private equity firm EQT. They also note the shutdown of Viva Translate, a speech-to-speech translation company that will open-source its tools as it winds down.

Highlights from SlatorCon Silicon Valley 2024

Slator- Language Industry Intelligence

On September 5, 2024, more than 150 language industry and technology leaders gathered at Hotel Nia in Menlo Park, Silicon Valley.

The event offered a friendly and relaxed environment, encouraging networking and reconnections among participants. Attendees from over a dozen countries and four continents emphasized the importance of in-person Slator events in addition to virtual ones. The expo hall was also buzzing with activity.

Esther Bond, Head of Advisory at Slator, kicked off the event with a warm welcome, outlining the day's presentations and panels, and encouraging delegates to network and engage with each other.

Key Takeaways from SlatorCon Silicon Valley 2024

Florian Faes, Managing Director of Slator, opened the sessions by presenting key insights from Slator’s latest research on the language industry's current state. He discussed practical applications of large language models (LLMs) in localization workflows and shared predictions for the next few years.

RWS took the stage for the first presentation, with Vasagi Kothandapani and Mark Lawyer discussing the diversification of services into AI solutions. They emphasized the role of content as a driving force for digital transformation, business innovation, enhancing customer experience, corporate growth, global engagement, and market evolution.

Key Takeaways from SlatorCon Silicon Valley 2024

The day's first panel, moderated by Esther Bond, focused on investment strategies.

Andrew Doane of K1 Investment Management and Aditya Govil from VSS Capital Partners explored the influence of AI on the language technology sector, with particular emphasis on the healthcare and B2B SaaS industries. They also discussed the role of private equity in the language tech space and shared insights on strategic considerations for investments and acquisitions.

Helena Batt, who oversees localization operations for the TED Conferences, took the podium next to provide unique insights on the organization’s implementation of AI dubbing for TED Talks. Among the technical challenges encountered, Batt mentioned preserving vocal characteristics and emotional nuance, and achieving seamless lip sync.

Betting on Technology

The Language AI Stack panel, moderated by Anna Wyndham, Slator's Head of Research, featured insights from Georg Ell of Phrase and Hameed Afssari of Uber. They discussed AI as a technology stack, focusing on the practical applications of large language models (LLMs) in localization, including machine translation (MT), workflow optimization, and managing linguistic assets.

A second technology panel, led by Florian Faes, explored the interpreting field. Oddmund Braaten from Interprefy, Fardad Zabetian from KUDO, and Jeremy Woan from CyraCom International shared their perspectives on how automation transforms interpreting services.

Another panel, moderated by Alex Edwards, Slator Senior Research Analyst, offered insights on localization systems integration, global 24/7 services, and enterprise program management. Panelists included Pavel Soukenik from Acolad, Nitin Singhal from SnapLogic, and Agustín Da Fieno Delucchi from Microsoft.

Silvio Picinini from eBay Localization delivered a thought-provoking presentation, exploring two scenarios: applying AI to existing localization processes or reimagining those processes entirely, and the potential outcomes of each approach.

Florian Faes concluded the event with closing remarks, inviting attendees to join SlatorCon Remote in November 2024 or meet in person again at SlatorCon London in 2025. More detailed follow-up coverage is forthcoming.

 Researchers Combine DeepL and GPT-4 to Automate (Research) Questionnaire Translation

In today's fast-paced research environment, the demand for accurate and swift translation of research questionnaires is higher than ever. With global collaborations becoming the norm, researchers are constantly on the lookout for efficient ways to break language barriers. Enter the world of AI, where tools like DeepL and GPT-4 are changing the game. But how exactly does combining these technologies automate research questionnaire translation, and why is it a big deal? Let's dive in.

Slator- Language Industry Intelligence

The Role of DeepL in Translation

Overview of DeepL

DeepL has quickly risen to prominence as one of the most reliable AI-powered translation tools available. Known for its impressive accuracy, DeepL has been widely adopted by professionals who need translations that go beyond the literal, capturing the essence and context of the source material.

Advantages of Using DeepL for Translation

The primary strength of DeepL lies in its ability to understand context. Unlike traditional translation tools that may deliver clunky or out-of-context translations, DeepL leverages deep learning algorithms to provide translations that are more natural and nuanced. This makes it particularly valuable in the field of research, where the precise meaning of each question is critical.

Understanding GPT-4

Introduction to GPT-4

GPT-4, the latest iteration in OpenAI's Generative Pre-trained Transformer series, is a language model designed to understand and generate human-like text. Its capacity to grasp complex large language model (LLM) and produce coherent responses makes it a powerful tool in various applications, including translation.

How GPT-4 Enhances Language Understanding

What sets GPT-4 apart is its ability to understand and generate text with a high degree of fluency. It can process a wide range of languages and adapt to different linguistic contexts, making it an invaluable partner in translation tasks. This adaptability is crucial when dealing with research questionnaires that often contain specialized terminology and nuanced language.

Combining DeepL and GPT-4 for Research Translation

Why Combine DeepL and GPT-4?

While both DeepL and GPT-4 are powerful on their own, combining them creates a synergy that enhances the overall translation process. DeepL's contextual accuracy pairs well with GPT-4's ability to generate coherent and contextually appropriate text, leading to translations that are both accurate and natural.

The Synergy Between DeepL and GPT-4

When used together, DeepL can provide the initial translation, which GPT-4 can then refine. This collaboration allows for more precise and culturally sensitive translations, which are particularly important in research, where misinterpretations can lead to skewed data.

Benefits of Automating Research Questionnaire Translation

Speed and Efficiency

One of the most significant advantages of automating research questionnaire translation is the speed at which it can be done. What used to take weeks or even months can now be accomplished in a matter of hours, allowing researchers to focus more on analysis and less on administrative tasks.

Cost-Effectiveness

Automating the translation process also cuts down on costs. By reducing the need for human translators, especially for initial drafts, organizations can allocate resources more effectively.

Improved Accuracy

AI-driven translation tools, especially when combined, can achieve a level of accuracy that minimizes the risk of errors. This is particularly important in research, where the integrity of data relies heavily on the clarity and precision of the questions asked.

Case Studies of Automated Translation in Research

Real-World Applications

Several organizations have already begun integrating AI into their research translation processes. For instance, a leading university in Europe used the DeepL and GPT-4 combination to translate questionnaires for a multi-country study on public health. The results were impressive, with over 90% accuracy in the translations, far surpassing traditional methods.

Success Stories

Another success story comes from a non-profit organization that conducts surveys in various languages. By automating their translation process, they not only sped up data collection but also improved the quality of their translations, leading to more reliable survey results.

Challenges in Automating Research Questionnaire Translation

Language Nuances

One of the primary challenges in automating translation is dealing with language nuances. AI tools must be able to understand idiomatic expressions, slang, and cultural references, which can be difficult for even the most advanced systems.

Contextual Understanding

Ensuring that the context is preserved during translation is another hurdle. Research questionnaires often contain complex ideas that need to be conveyed accurately in another language. Misinterpretation of these ideas can lead to data that is not only inaccurate but also unusable.

Ethical Considerations

As with any AI application, there are ethical considerations to keep in mind. Issues like data privacy, the potential for bias in translations, and the role of human oversight are all critical factors that need to be addressed.

Overcoming Challenges with AI Technology

Advances in AI for Better Translation

Recent advancements in AI are helping to overcome many of the challenges associated with automated translation. For example, improvements in natural language processing (NLP) are allowing AI systems to better understand context and nuances, leading to more accurate translations.

Role of Human Oversight

While AI can significantly enhance the translation process, the role of human oversight cannot be understated. Human translators can provide the necessary context and cultural understanding that AI might miss, ensuring that the final translation is both accurate and culturally appropriate.

Future of AI in Research Translation

Emerging Trends

The future of AI in research translation looks promising, with trends pointing towards even more sophisticated and reliable tools. As AI continues to evolve, we can expect to see greater integration of these technologies in research, leading to faster and more accurate data collection.

Potential Impact on Global Research

The global impact of AI-driven research translation cannot be overstated. By breaking down language barriers, AI has the potential to democratize research, making it more accessible to non-English speaking populations and fostering greater collaboration across borders.

Conclusion

In conclusion, the combination of DeepL and GPT-4 represents a significant advancement in the field of research translation. By automating the translation process, researchers can save time, reduce costs, and improve the accuracy of their data. While challenges remain, the future looks bright, with AI poised to play an increasingly important role in global research.

Create a Unique Blog with Anchor Tags: The Impact of Emotional Context on LLM Translation Quality

Slator- Language Industry Intelligence

In an increasingly globalized world, translation is more important than ever. With the rise of Large Language Models (LLMs), translation has become more efficient and accessible. However, one crucial aspect that often gets overlooked is the role of emotion in translation. Emotion isn’t just a secondary component of language; it’s a vital part of communication that can significantly affect the quality of translations produced by LLMs.

The Growing Importance of Translation in the Digital Age

The internet has connected people from different parts of the world like never before, and translation services are at the heart of this connection. Whether it's translating a website, a marketing campaign, or a casual conversation, accurate translation is essential for clear communication across cultures.

What are Large Language Models (LLMs)?

LLMs are advanced machine learning models designed to understand and generate human language. They are the backbone of modern translation tools, enabling users to convert text from one language to another with unprecedented ease.

How LLMs Function in Translation

LLMs work by analyzing vast amounts of text data to learn the intricacies of language. They use this knowledge to predict and generate text that is coherent and contextually appropriate. However, while LLMs are excellent at handling syntax and grammar, they often struggle with the more nuanced aspects of language, such as emotion.

The Challenges of Accurate Translation

One of the biggest challenges in translation is capturing the true meaning of a text, especially when emotions are involved. Emotions can change the meaning of words and phrases, and without understanding the emotional context, translations can easily miss the mark.

The Role of Emotion in Communication

Emotion plays a critical role in how we communicate. It influences not just the words we choose but also how those words are interpreted by others.

How Emotion Influences Language Understanding

When we communicate, we do more than just exchange words; we convey emotions, intentions, and subtle nuances that can change the meaning of what we say. For example, the phrase "I'm fine" can mean different things depending on the speaker's tone and emotional state.

The Impact of Emotional Context in Translation

When translating text, understanding the emotional context is crucial. A phrase that might be neutral in one language could be highly emotional in another. Without accounting for this, LLMs may produce translations that are technically accurate but emotionally off-base.

The Connection Between Emotion and Translation Quality

Emotional context is not just a nice to have in translation; it’s essential for accuracy and effectiveness.

Why Emotional Context Matters in Translation

Translation is not just about converting words from one language to another; it's about conveying the same meaning and feeling. Emotions add depth and authenticity to communication; without them, speech translations can be flat or misleading.

Misinterpretations in Emotion-Laden Texts

When emotional context is ignored, the results can be disastrous. A translation that doesn’t capture the intended emotion can lead to misunderstandings, offend readers, or even change the meaning of the original message entirely.

Examples of Emotional Context Misunderstandings

Consider a simple phrase like "Thank you." Depending on the context, it could be sincere, sarcastic, or even dismissive. A translation that fails to recognize these emotional cues could easily misinterpret the speaker's intent.

Enhancing LLMs with Emotional Context

To improve the quality of translations, LLMs need to be trained to understand and replicate emotional context.

How Emotional Data Can Improve Accuracy

Incorporating emotional data into LLM training can significantly enhance their ability to produce accurate translations. By learning to recognize and replicate emotional nuances, LLMs can generate translations that are not just accurate in terms of language but also terms of emotion.

Training LLMs with Emotional Nuances

Training LLMs to understand emotion involves feeding them text data that includes emotional context. This can be done by tagging text with emotional labels or by using advanced techniques like sentiment analysis to help the model learn to recognize emotional cues.

Real-World Applications of Emotionally Enhanced Translation

Emotionally intelligent LLMs can be game-changers in various fields where translation is crucial.

Business Communication

In business, clear and accurate communication is key, and emotion plays a big role in this.

Marketing and Customer Engagement

Marketing campaigns rely heavily on emotional appeal to connect with customers. An emotionally intelligent translation can ensure that the intended message resonates with the target audience, regardless of the language.

International Negotiations

Negotiations involve not just exchanging information but also understanding the emotional undercurrents of the conversation. A translation that captures these subtleties can make a big difference in the outcome of negotiations.

Media and Entertainment

The media and entertainment industry also relies on translation to reach global audiences.

Film Subtitles and Dubbing

Subtitles and dubbing are more than just translating dialogue; they are about conveying the same emotions that the original actors expressed. Emotionally aware translations can make foreign films more relatable to global audiences.

Literature and Script Translation

When translating literature or scripts, capturing the emotional depth of the original text is crucial. Emotionally intelligent LLMs can help preserve the author's voice and the emotional impact of the story.

Healthcare and Legal Communication

In fields like healthcare and law, where the stakes are high, accurate translation is critical.

Patient-Doctor Communication

In healthcare, miscommunication can have serious consequences. An emotionally intelligent translation can help ensure that patients and doctors understand each other fully, reducing the risk of errors.

Legal Document Translation

Legal documents are often complex and filled with nuances. A translation that accurately conveys the emotional weight of these documents can be crucial in legal proceedings.

Conclusion

The future of translation lies in emotionally intelligent LLMs. As these models continue to evolve, their ability to understand and replicate emotional context will become increasingly important. This will not only improve the accuracy of translations but also make them more human-like, enabling better communication across cultures.