Zhao Bin, CEO of SoundNet: "SoundNet has achieved broadcasting level 4K ultra high definition real-time interactive experience for the first time in the RTE industry"

On October 25th, it was reported that at the 9th RTE2023 Real Time Internet Conference, Zhao Bin, founder and CEO of SoundNet, announced that "SoundNet has achieved a broadcast grade 4K ultra high-definition real-time interactive experience for the first time in the RTE industry." He pointed out that "4K not only has a resolution dimension, but also the dynamic range of brightness intensity, the authenticity and saturation of colors, and the delicacy of color gradient are all indispensable parts of a true ultra high-definition experienceZhao Bin believes that the emotional and emotional value created by virtual humans and AIGC may solve the imbalance between social supply and demand in the future and lead to changes in interaction objects

On October 25th, it was reported that at the 9th RTE2023 Real Time Internet Conference, Zhao Bin, founder and CEO of SoundNet, announced that "SoundNet has achieved a broadcast grade 4K ultra high-definition real-time interactive experience for the first time in the RTE industry." He pointed out that "4K not only has a resolution dimension, but also the dynamic range of brightness intensity, the authenticity and saturation of colors, and the delicacy of color gradient are all indispensable parts of a true ultra high-definition experience

Zhao Bin believes that the emotional and emotional value created by virtual humans and AIGC may solve the imbalance between social supply and demand in the future and lead to changes in interaction objects. On the other hand, with the further strengthening of AIGC capabilities, more and more application based development will be replaced by AIGC, and the form of API+AI will greatly improve application development efficiency.

Large models are rapidly developing, but the interactive interface is still limited to text chat interaction. Zhao Bin mentioned that through investment and refinement over the past year, SoundNet has launched the industry's first AIGC-RTC capability module, which can be combined with any large model platform to achieve lower latency, more natural, and more immersive real-time voice conversations, providing new gameplay and opportunities for various scenarios such as enterprise collaboration, social networking, live streaming, and gaming.

According to Sun Yurun, Partner and Vice President of Customer Success at SoundNet, "according to data from a leading Southeast Asian pan entertainment customer at SoundNet, when a video is upgraded from standard definition to high-definition and stays for more than 30 seconds, the number of viewers increases by 19%, and the average viewing time per viewer increases by 30%. At the same time, the audience reward rate has significantly increased. A well-known domestic social networking app has upgraded from CDN to SoundNet's high-speed live streaming, and the gift rate has increased by 12.3%

Zhong Sheng, Chief Scientist and CTO of SoundNet, pointed out that the most typical event in the past five years is the breakthrough of big models. From Transformer in 2017 to ChatGPT-4 today, the development of artificial intelligence has made the trend of information dissemination and consumer intelligence increasingly apparent, and the trend and threat of everything being taken over by centralized AGI are also becoming increasingly apparent. While the big model brings development opportunities, it also brings problems such as rapid growth of computing demand, limited computing power by energy supply, insufficient data resources of the big model, and rapid growth of storage demand. In the future, in addition to responsible AI, distributed real-time intelligence on the end and edge will become an important technological means for fair distribution of value and an effective way to mitigate the threat of centralized AGI to humanity. This is also destined to become a new technological development trend.

Zhong Sheng believes that AGI will enter real-time interaction, enabling everyone to be separated, helping to replicate famous teachers, internet celebrities, and even ordinary people to enrich their experience and alleviate the bottleneck of time scarcity through AI separation in application scenarios. In addition, he predicts that high-definition real-time interactive capabilities with end-to-end real-time intelligence will become a new trend and competitive focus.

Zhong Sheng introduced that real-time high-definition requires a lot of real-time AI on the end. At the demand level of LowLevelVisionandAudio, the SDK4.1. x and 4.2. x versions of SoundNet can already support low-level computer visual processing and high-quality computer auditory processing capabilities for 1080P/4K videos, such as superresolution, virtual background, perceptual encoding, noise reduction, and echo cancellation. At the HighLevelVisionnandAudio level, the sound network has partially achieved the ability to understand and reconstruct objects, sounds, and scenes, including surface capture, motion capture, emotion computing, object recognition, and scene reconstruction, which can greatly improve user experience in various application scenarios. (One Orange)