According to Dujia, Soul will launch "Loxinity (洛希无限)", an integrated AI audio-video product built for interaction scenarios, comprising the speech synthesis model Loxi Speech and the video generation model Loxi Avatar, positioned as a real-time interactive "human-like" voice and multimodal product.
Unlike traditional TTS on the market aimed at dubbing and short videos, Loxinity targets real-time conversation scenarios at its core, placing greater emphasis on paralinguistic details such as tonal pauses and emotional fluctuations. The product supports two modes, single-person dubbing and multi-person dialogue, comes with tools such as voice cloning and voice design, and offers API access. On the video side, users can upload images to generate digital human avatars, and the platform also offers preset characters in realistic, anime, and fantasy styles, supporting real-time video calls.
According to hands-on testing, the Loxinity interface has integrated visual configuration options such as "add paralinguistic features" and "add emotion", allowing users to finely control the emotional expression of the voice. The product adopts two billing models, credits and subscription plans, and is open to B2B customers.
Soul has previously conducted multiple AI social experiments within its app, including AI group chat hosts and AI companions, accumulating massive amounts of real human conversation data, which has become its core moat in polishing "human-like" interaction technology.
Analysts believe that, against the backdrop of giants crowding the general large model track, Soul has chosen the differentiated track of emotional interaction, but when capabilities validated within its own app are delivered to B2B clients, concurrent capacity, real-time latency, and project costs remain hard thresholds, and its commercialization potential remains to be tested by the market.