logologo

Soul App's Three New Features Unveiled at WAIC, Epitomizing the Product Philosophy of “Synergic Model-application Interplay”

On July 4th, the 2024 World Artificial Intelligence Conference (WAIC 2024) officially opened. With this year's theme being "Governing AI for Good and for All", it concentrated on the three major sections of core technology, intelligent terminals, and application empowerment.

As a leading social platform which incorporates AI technology, Soul App was invited to participate in the exhibition and displayed a couple of its new features in AI field, such as “Digital Twin”,“Werewolf Awakening”, and “EchoVerse”.

图片_20240708174351.jpg

Three new functions are unveiled, demonstrating multimodal interaction capabilities

At the WAIC 2024 event venue, Soul presented its booth under the theme of "Human Touch, Digital Heart" and particularly demonstrated three new features, namely "Digital Twin", "Werewolf Awakening", and "EchoVerse", which constitutes Soul's latest practice in exploring the profound integration of "AIGC + Social".

Among them, "Digital Twin" is dedicated to assisting users in creating virtual digital humans for efficient socializing. Users can authorize the platform to set the image and characteristics of the digital twin based on chat records and posting content in Soul’s public domain scenes. In rich dimensions such as personas, images, and voices, the digital twin can achieve the effect of replicating a real person to the greatest extent.

Adhering to interest-based socializing, the Soul platform does not support the utilization of real human profile pictures. Users create virtual images through the avatar system for interaction in the digital space. Currently, the launch of the "Digital Twin" with private chat assistance capabilities not only helps users create a more ideal "other self", but also enables more personalized and diverse intelligent reply recommendations, facilitating users in breaking the ice in socializing and enhancing social communication efficiency when assisting with persona building and cognitive decision-making.

Also, the introduction of AI entities in the game interaction scenario "Werewolf Awakening" prominently showcases the multimodal interaction capabilities of the large model. In this scenario, users can opt for any combination mode of AI and real-person to initiate the interaction and engage in a real game showdown with AI entities possessing autonomous reasoning, speech, and "disguise" capabilities. The "AI game companion" can also assist players in quickly adapting to the relatively high threshold and complex gameplay of the werewolf game, easily initiating communication, and obtaining an enjoyable experience in a more immersive and real-time interaction.

While integrating AI capabilities into in-site scenarios, Soul has launched an independent new feature, "EchoVerse". This feature is positioned as an AI social platform where users can engage in immersive real-time communication with virtual characters and customize the personas of the characters according to their preferences to achieve different conversation styles. The image of the character can be generated through text description or by uploading a desired picture. The platform offers multiple basic timbres, and users can independently create and fuse the exclusive voice of the character to realize multimodal interaction.

"Synergic Model-application Interplay" practice, from social improvement to human-computer interaction new experience

In fact, as a platform that began contemplating the application of AI in the social field relatively early, Soul has embarked on the exploration of underlying technologies and application implementation around AI since its launch in 2016.

For instance, by launching the intelligent recommendation engine "Ling Xi" based on the full-scene portrait in the user station, the platform helps users find people with the same interests with a decentralized mechanism, and equally realizes the establishment of multiple relationships and instant emotional feedback, effectively improving the user experience. In addition, the platform's NAWA engine enables users to create personalized avatars and scenes for immersive interaction.

In 2020, Soul initiated systematic research and development work on AIGC and has accumulated cutting-edge capabilities in intelligent dialogue, image generation, and voice technology, such as voice generation, music generation, voice animation, etc. In 2023, Soul launched its self-developed large language model, SoulX, which possesses capabilities such as prompt-driven, conditionally controllable generation, context understanding, and multimodal understanding, and can achieve emotional and congenial interaction.

In 2024, Soul's voice generation large model was launched, and the self-developed voice large model was officially upgraded. At this stage, Soul's voice large model encompasses voice generation large model, voice recognition large model, voice dialogue large model, music generation large model, and is capable of supporting functions such as real timbre generation, voice DIY, multi-language switching, and multi-emotional real-person-like real-time dialogue.

Nowadays, the achieved multimodal interaction capabilities of Soul have been integrated into specific application scenarios including Soul's "AI Godan", Werewolf Awakening, Digital Twin, and EchoVerse, further realizing the enhancement and expansion in aspects such as interaction efficiency, interaction quality, interaction experience, and interaction objects.

"Users' willingness to continuously engage in conversations and interactions with AI has already proved their recognition of the experience provided by the platform. It also indicates the sustainability of Soul's insistence on the direction of "Synergic Model-application Interplay", said Tao Ming, the CTO of Soul App.

AI is like finding the right hammer for the nail, Tao Ming said. "Soul is one of the most popular Internet platforms among Chinese young people. It is a natural application scene and a traffic entrance. Our advantage is we can identify the actual needs of users in the scene and the application of the large model can also obtain feedback from users immediately."