From Hardware to AI: Evolution of Language Accessibility in Events

After years of fully remote conferences, on-site events are again on the rise, with in-person attendance up 23% compared to pre-pandemic levels. And while this request is coming mainly from attendees themselves, longing for face-to-face interactions, it also opens new questions for event organizers: 

• How can I expand the global reach of my tradeshows? 
• How to make sure my events are fully language-accessible for international speakers and audiences? 
• How expensive will this be? And what about the ROI?  
• How can AI translation tools help me? 

It’s a fact: over the last few years, the development of AI technology has transformed the congress and conferencing world. Efficiency, engagement, and productivity are the typical driving factors behind its adoption, but we’re also seeing a new use case: accessibility of event content for speakers of different languages. In other words, AI-powered live translation. But let’s rewind and see how we got here. 

In the past, event organizers needed a lot of hardware to make an in-person event multilingual, and that’s not counting the cost of booking on-site interpreters for the event. As interpreters joined events and interpreted on-site, organizers had to factor in things like headsets, bulky battery packs and other components that were needed, such as an isolation booth for the interpreter. Then, with the advent of virtual and hybrid conferences, new developments shaped the event tech industry with the introduction of RSI (Remote Simultaneous Interpretation) and the option of having interpreters connect remotely.  

Today, thanks to AI Speech Translation technology, we have an innovative solution that allows participants to use their own personal devices to access live-translated audio and captions – all they need to do is simply scan a QR code, plug in their headphones and instantly access simultaneous AI interpretation. A long way away from messy headsets and battery packs, and a huge boost for language accessibility.

Before considering the benefits of using AI speech translation in your in-person events – and how to choose between AI and human interpreters – let’s address the obvious: ‘why should I make my event multilingual in the first place?’   

Making your events multilingual is crucial for increasing both the reach and the engagement of your event, regardless of the setting—on-site, remote, or hybrid. By incorporating simultaneous interpretation into your meetings, powered either by AI or conference interpreters, you eliminate any limitations regarding speaker selection. This means you can choose the best ones, regardless of their linguistic background, while ensuring that your audience fully understands the content, asks questions, and actively participates in the discussions in their own language.  

Therefore, by offering both speech/audio translation and closed captioning, event managers guarantee a language-accessible experience for all attendees, regardless of their language background. But what about the full event ROI?  

This is where AI gets a big green check in the ‘pros’ column. Rather than being charged for each individual language you need, most AI speech translators will have an hourly cost that includes all languages. So, whether you’re trying to reach one new region with your event or ten, the price is the same. 

Making your content multilingual offers multiple other benefits:  

• Scalability and adaptability: it boasts high scalability, effortlessly managing large volumes of simultaneous translation requests across multiple languages. 
• Cost-efficient: AI speech translation offers remarkable cost-efficiency, with minimal ongoing or additional costs regardless of the number of languages requested. 
• High performance: the exceptional performance of AI speech translation, coupled with its accuracy, speed, and minimal setup requirements, instills confidence in its ability to deliver and integrate seamlessly into any event environment.

Beyond the opportunities, AI speech translation is still a new technology – and it’s understandable to have reservations around using it. In a recent webinar, we asked our attendees about theirs, and while a solid 21% of our audience shared they won’t have any reservation about testing AI, we also found out that: 

• The majority (41%) is worried AI won’t capture context, specific terminology, or cultural nuances, 
• 19% is worried about AI not being accurate enough (e.g. capturing regional accents), 
• 9% is concerned that AI won’t be suitable for conversations.

How do AI translation apps stand against these reservations? The answer is made up of two elements – how to correctly evaluate machine interpreting quality, and how products are evolving. 

Starting with the first point, it’s important to note that there is no universally accepted definition of ‘translation quality’ applicable to both human and AI translations. Numerous variables influence how users perceive quality, making the evaluation of data a real challenge. 

The components to consider include: 

• Accuracy: The capacity to convey the semantic essence of the original message without distortion or contextual loss. 
• Fluency: This pertains to a translation’s continuity, smoothness, and its ability to capture colloquialisms. 
• Latency: The delay between the speaker’s statement and the listener hearing the translated voice, measured in seconds or the number of words. It’s worth noting that latency is not exclusive to machine translation; even conference interpreters need several seconds to comprehend what’s being said before they start interpreting. 

Machine interpreting scores extremely high in terms of accuracy and fluency, thanks to the growing sophistication of text-to-text machine translators like Microsoft Azure or DeepL, which are foundational to speech-to-speech architecture. Latency presents however a different challenge, as the system may require a specific context to produce a coherent translation, depending on the language combination or the speaker’s style. 

Blind tests have shown that KUDO has much lower latency compared to competitors, and we are continuously working on technology enhancements to reduce latency and improve overall quality. For example: 

• Speedometer: A proprietary feature now integrated into all KUDO AI meetings; the Speedometer addresses the challenge of compromised translation quality due to rapid speech. It provides real-time feedback on the speaker’s pace of speech, monitoring crucial attributes like translation difficulty, sound quality, and speech density, and sends an alert if the pace is too fast for simultaneous interpretation. 
• Custom Glossary: Enhancing user experience further, KUDO AI users can now make their AI-translated meetings more accurate with the Custom Glossary feature. Users can pre-enter a list of specific subject matter terms, words, names, products, places, and acronyms for correct translation by KUDO AI. This is particularly beneficial for events such as medical conferences and product launches. 

It’s important to clarify that AI is not in competition with human conference interpreters. The overall experience is different, and each serves different use cases. In fact, they can work in combination and at KUDO we have quite a few clients who use both. 

AI is an increasingly popular choice for one-to-many presentations such as keynotes, company-wide meetings including all-hands, and leadership meetings. It’s also a great alternative when sourcing a conference interpreter for less common languages is challenging. 

AI also offers the advantage of full flexibility, being available 24/7 for on-demand meetings without the need to book or prepare an interpreter in advance. With KUDO, we can also facilitate permanent setups, for example, imagine if people could access language interpretation anytime in a room where lectures or press conferences are regularly hosted. 

Compared to human interpretation, AI translation is certainly more affordable, particularly when offering multiple language combinations at your event, since AI imposes no restrictions on how many languages can be translated simultaneously. Thus, AI becomes a cost-effective solution when the budget cannot accommodate the expense of human interpreters. 

On the other hand, human interpreters are the preferred choice for highly interactive meetings as it accommodates two-way communication, whereas AI is just one way. Human interpreters also have full control over the terminology and context of your meeting and can understand and replicate the nuances in expressions and intonation, providing your attendees with an enhanced multilingual experience.

To summarize, the evolution of interpretation capabilities for live events is undeniably impressive, and it’s continuously advancing. AI translation technologies mark a pivotal moment for on-site events, offering unparalleled performance particularly suited for large-scale gatherings like congresses, press conferences and tradeshows. 

The real differentiator with AI here is scalability and cost-efficiency. With no additional expenses for multilingual support, event organizers can significantly expand their event’s global reach, while optimizing return on investment.  

Whether opting for AI-driven solutions or traditional human interpretation, the imperative remains the same: embracing multilingualism and ensure language accessibility is paramount for enhancing audience engagement and broadening the impact of any event. From interpreter booths to cutting-edge AI speech translation, the ability to create meaningful connections across linguistic barriers is constantly evolving.