Soul will launch Loxinity, a self-developed product with voice, vision and multimodal interaction capabilities.

Exclusive by ReadJia, Soul is set to launch "Loxinity", a speech synthesis product tailored for interactive scenarios with extremely lifelike human voice experience, different from the TTS products for dubbing and short videos on the market, this model is built from the ground up for real-time interactive scenarios, with a focus on highly authentic human-like voice performance. It supports two modes: single-person dubbing and multi-person dialogue, comes with supporting tools such as timbre cloning and timbre design, and also provides open API access. Loxinity encompasses the speech synthesis model Loxi Speech and the video generation model Loxi Avatar.

At present, compared with traditional speech models that only pursue standard pronunciation, Loxinity pays more attention to paralinguistic details such as tone pauses and emotional fluctuations in conversations, to support continuous voice interaction in social scenarios.

The product's mission is: To make AI interaction more natural — shortening the distance between humans and intelligence through voice, visual and multimodal technologies. With such an ambitious positioning, it is not ruled out that this product will become an important attempt by Soul in the AI voice scenario beyond its core social business.

Soul brings AI voice, real-time video call and digital human to the market as a packaged solution

Actual test of Soul's AI product Loxinity by ReadJia shows that this product is based on the speech synthesis model Loxi Speech and the video generation model Loxi Avatar. It is not just a single speech synthesis tool, but an integrated AI audio and video solution for real-time interaction, with three core sections: speech synthesis, real-time video call and digital human.

The speech synthesis module supports single-person dubbing and multi-person dialogue, comes with timbre cloning and timbre design functions, and also allows adding paralinguistic elements, pinyin and emotion labels. Its optimization focus is on dialogue scenarios, emphasizing the restoration of pauses and tone fluctuations in real human speech. It also provides open API access, adopting two billing modes: points-based and subscription packages.

The digital human and real-time video call section is another key focus of this product. Users can upload images to customize their digital human avatars, input text scripts or upload audio to generate corresponding digital human videos.

The platform presets a variety of roles in realistic, anime, fantasy and other styles, for building AI virtual characters that support real-time dialogue, and a large number of user-created cases such as interviews and lectures are displayed on the page.

Overall, Loxinity avoids the crowded general large model track and focuses on interactive audio and video AI. Many similar digital human tools have problems such as heavy robotic reading tone and out-of-sync audio and video. Loxinity chooses to natively integrate speech and digital human driving, so that the speech capability can support continuous dialogue.

Of course, there are still uncertainties. For this type of real-time interactive product, the core bottlenecks lie in long conversation stability and end-to-end latency. After leaving Soul's internal social scenarios and being delivered to external B-end customers, its concurrent processing capability, usage cost, and naturalness of digital human micro-expressions still need to be verified by more landing projects. How much market recognition this AI interactive product from a social platform can ultimately gain remains to be seen.

From in-app social scenarios to external output: Soul bets on interactive AI

The core demand of young people for AI is not text Q&A, but immersive interaction with emotions that is close to real human experience. Long before the launch of Loxinity, Soul had carried out multiple rounds of pilot attempts within its platform.

In 2022, Soul launched the AI chatbot "AI Godan", as the first-generation AI companion role, focusing on daily-life conversations and active emotional care, no longer limited to Q&A-style interaction. In the same period, it launched the NAWA virtual avatar engine, allowing users to create their own virtual avatars, embed AI dialogues into virtual character carriers, and build virtual social scenarios. In 2023, Soul officially released its self-developed emotion-focused vertical large model SoulX, positioned as an emotion-aware interaction base, instead of following the trend of developing general large models, it focuses on strengthening the AI's capabilities of understanding emotions, maintaining consistent character settings, and long-context memory.

Relying on the SoulX base, a series of AI social products have been launched one after another. The platform launched AI painting and AI anime singer functions, allowing users to create virtual avatars and AI human voices; it launched an end-to-end full-duplex speech model, implemented AI hosts in group chat parties, supporting real-time interjection and natural interruption, reproducing the dialogue rhythm of real multi-person casual chats, and getting rid of mechanical interaction. Scenarios such as in-app Werewolf game and virtual companion are all connected to this speech capability, realizing long-turn continuous voice dialogues. These in-app scenarios have provided a testing ground for continuous iteration and polishing of this "human-like" interactive technology.

The general large model track is already crowded with tech giants. Soul did not compete head-on, but chose the emotion interaction track. The massive sample of real human dialogues accumulated from years of social business operations is its core barrier for polishing paralinguistic features and natural dialogue rhythm.

However, the challenges are equally clear: when the interaction capabilities verified to be feasible in in-app scenarios are delivered to B-end customers externally, concurrent load capacity, real-time latency, and project cost are all rigid thresholds. The commercialization potential of this AI capability still needs to be tested by market landing cases. #Soul#Soul launches AI product#Soul launches Loxinity