Breakthrough AI Translator Delivers Real-Time Simultaneous Interpretation in Online Conferences
Until recently, the idea of real-time simultaneous interpretation in online conferences would have seemed like science fiction. However, thanks to the advancement of artificial intelligence (AI), a breakthrough has occurred in the form of an AI translator that can provide instantaneous translations during video conferences. Kudo, a company that connects interpreters with corporate clients, has incorporated this technology into its services, allowing participants in online conferences to hear translations as if they had a live interpreter present.
In a recent demonstration for EL PAÍS, Tzachi Levy, Kudo’s product manager, spoke in English while being simultaneously interpreted into Spanish. Although the AI-generated voice sounded slightly robotic with a slight delay compared to a human translation, the results were still impressive. While a human interpreter typically has a delay of five to seven seconds, the AI system experienced a delay of around 10 seconds. Despite this, Kudo’s technology is constantly being improved and has already attracted 20 corporate clients. It is currently integrated with Microsoft Teams, a popular video conferencing platform in the corporate world.
It is worth noting that in situations where 100% accuracy is required, human interpreters are still considered the best option. For instance, in European Parliament sessions, AI systems are unlikely to be used due to the complexity of the context. Nevertheless, there are numerous scenarios where AI translation can be effective, especially in smaller meetings where interpreters may not be readily available.
According to Levy, the progress in AI translation has been remarkable, surpassing initial expectations by achieving in months what was previously predicted to take years. He estimates that within the next year, AI could accurately provide simultaneous translations in 90% of common situations. However, there are still challenges to overcome, particularly in interpreting contextual cues and reading between the lines. Despite this, large language models like ChatGPT are contributing to advancements in this area.
Claudio Fantinuoli, head of technology and creator of Kudo’s automatic translation tool, believes that there is no conflict between AI and human training. In fact, he emphasizes that human interpreters will always offer superior quality. Fantinuoli encourages aspiring interpreters to embrace AI technology and strive to excel in their profession. He also mentions the possibility of integrating the speaker’s own voice into the interpretation, a feature that is expected to be available in the near future.
The future of AI translation looks promising, with Meta recently launching a multimodal translation model capable of performing various translation tasks for up to 100 languages. This inclusive model can decipher the different languages at play in a sentence and provide accurate translations. While small errors still exist, the tool shows great potential and is available for free in beta.
In conclusion, AI translation has come a long way in delivering real-time simultaneous interpretation in online conferences. While it may not be perfect yet, the advancements in AI technology are rapidly closing the gap between human and AI translation. As the field continues to evolve, we can expect even greater accuracy and usability in the near future, making global communication more seamless and accessible to all.