Core Viewpoint - Soul is set to internally test its self-developed end-to-end full-duplex voice call model, which will be applied in various interactive scenarios such as real-time communication with virtual humans and AI matching [1] Group 1: Development Timeline - In 2020, Soul initiated the research and development of AIGC (Artificial Intelligence Generated Content) technology [1] - In 2023, Soul launched its self-developed Soul X, followed by the rollout of various large model capabilities including voice generation, voice call, and music generation models [1] - By 2024, Soul's AI large model capabilities will be upgraded to a multi-modal end-to-end model, supporting text dialogue, voice calls, multi-language, multi-modal understanding, and realistic human-like interactions [1]

Core Insights - The 2025 World Artificial Intelligence Conference (WAIC 2025) is being held in Shanghai from July 26 to 28, showcasing advancements in AI technology [1] - Soul App has introduced its upgraded end-to-end full-duplex voice call model, demonstrating its application in multi-user voice interaction scenarios [1][4] Company Developments - Soul App's CTO emphasized the integration of AI into social networks, aiming to enhance user experience and foster a sense of belonging [3] - The platform showcased "AI social practice," highlighting the dual roles of AI as a Companion for emotional support and a Connector for facilitating real friendships [3] Technological Advancements - The newly upgraded full-duplex voice call model redefines interaction paradigms by eliminating traditional voice activity detection mechanisms, allowing AI to autonomously manage conversation flow [4] - Upcoming internal testing of the full-duplex call model will support various interaction scenarios, including real-time communication with virtual personas and AI matching [4] - The event also featured advancements in real-time video generation, enabling users to visualize AI personas and deepen relational connections [4]

Core Viewpoint - The article discusses the innovative solutions presented by Soul App at WAIC 2025, focusing on the integration of AI in social interactions, emphasizing the development of an end-to-end full-duplex voice call model that enhances user experience and emotional connection in social settings [2][4]. Group 1: AI Social Interaction Innovations - Soul App showcased its latest self-developed full-duplex voice call model, which allows for more immersive and intelligent emotional support and resonance in social interactions [2][4]. - The platform aims to redefine social paradigms by integrating AI into social networks, enhancing user experiences and fostering a sense of belonging [2][4][12]. - The full-duplex model eliminates traditional voice interaction limitations, enabling AI to autonomously manage conversation flow and enhance user engagement in group settings [4][5]. Group 2: User Experience and Emotional Value - The AI's role as a "Companion" and "Connector" allows it to provide personalized emotional support and facilitate real friendships among users [4][5]. - Soul's approach to social interaction emphasizes the importance of emotional value and information flow, moving beyond traditional tool-like interactions to a more genuine emotional coexistence [5][14]. - The platform's ability to capture and respond to users' emotional needs, such as loneliness and security, positions it to offer long-term emotional value through AI [14][15]. Group 3: Technological Advancements and Future Directions - Soul has been actively developing AI capabilities since its launch in 2016, focusing on areas such as intelligent dialogue, voice, and 3D virtual personas [12][13]. - The introduction of multi-modal end-to-end models in 2024 supports various interaction types, including text, voice, and real-time video, enhancing the depth of user relationships [12][8]. - The integration of AI into social experiences is expected to create new market opportunities and product paradigms, expanding the scope of social interactions and digital identity recognition [16].