Abu Dhabi Media and Chinese internet solutions provider Sogou Inc teamed up to develop an AI news anchor to present news on ADM’s channels. The AI news anchor incorporates industry-leading algorithms and the latest advances in speech synthesis, image detection and deep learning.
Wang Yanfeng, GM of Sogou’s Voice Interaction Technology Centre, tells BroadcastPro ME what the AI anchor can do at present and how it can free up time for human anchors.
What elements of a news anchor’s job can the AI currently do?
The AI news anchor incorporates the latest advances in speech synthesis, image detection and prediction capabilities, and can simulate natural speech and expressions by integrating various language-centric technologies. The AI anchor can replicate the in-studio news reporting experience.
What are the advantages of an AI anchor over a human one?
We believe that human anchors and virtual anchors serve different purposes in a news organisation. We see the AI news anchor as a complement to the work of a human broadcaster. AI anchors won’t replace the human connection that human news reporters have, but the technology will improve the efficiency of standard news broadcasts and free up human anchors to spend more time interviewing guests, reporting on issues outside of the studio and producing better quality content.
Can the AI anchor tackle anything that requires analysis, opinion or interaction with people?
The AI anchor generates news reporting by generating life-like expressions based on textual inputs. Multi-round human-machine interaction is a critical research area for the Sogou R&D team. In the not-too-distant future, we hope to be able to leverage Sogou’s Natural Language Processing (NLP) technology to deliver natural human-machine interactions tailored to support various scenarios. Across the entire industry, the technology required to conduct interviews or analysis is still at an early stage in its development and requires further research.
What technologies have gone into creating this news anchor – just machine learning and AI, or more? Essentially, what does it entail to create a virtual anchor?
The AI news anchor simulates natural speech and expressions by integrating various language-centric technologies. It incorporates the latest advances in speech synthesis, image detection and prediction capabilities, as well as complex algorithms, to generate life-like expressions.
Textual input can be transformed into corresponding lip movements, providing users with a highly customisable interactive experience.
Language lies at the core of Sogou’s AI platform and strategy. With a focus on natural language processing and machine learning, Sogou has developed unique capabilities in real-time audio and video synthesis.
What challenges do you face in creating such an anchor?
The pronunciation and expression of Arabic news anchors are different from English and Chinese news anchors, so a lot of work will go into developing an Arabic AI news anchor that can replicate a real-life news broadcaster. For example, there are several factors, such as the Arabic writing system, vowel recognition and the gender, number and case of Arabic words, that pose challenges to Arabic speech synthesis.
What can you do to make the anchor more sophisticated than those currently in use in China?
The ability of the AI anchor to replicate human news anchors is highly dependent on advances in the core technologies on which it has been developed. These include speech synthesis, natural language processing, image detection and prediction capabilities and AI technologies.
Sogou is already a leader in these technologies. At the core of AI are algorithms and big data. By combining our multi-language datasets, which are derived from Sogou’s Chinese-language input software and its AI R&D team’s expertise in algorithms, with Abu Dhabi Media’s position in the Arab media world, we are confident we can deliver a high-quality product and user experience.
We will continue to research and develop language-centric technologies that advance natural human-machine interaction, and to upgrade the AI anchor so that it can provide an increasingly realistic and seamless experience. We look forward to working with Abu Dhabi Media Company to co-develop an Arabic-speaking and English-speaking AI news anchor.
“AI anchors won’t replace the human connection that human news reporters have, but the technology will improve the efficiency of standard news broadcasts,” Wang Yanfeng, GM, Sogou’s Voice Interaction Technology Centre.
Will virtual anchors become the future of news TV?
AI has the potential to revolutionise the way we think about how humans interact with machines. Creating a more realistic virtual anchor will facilitate more seamless interactions and enable this technology to become an integral part of everyday life. There are also other potential applications for the AI anchor in industries such as education, healthcare and entertainment.