From Soul App to Industry: Soul at HUAWEI CONNECT 2026

During HUAWEI CONNECT 2026, Soul, the AI ecosystem company behind the Chinese AI social platform Soul App, jointly launched a solution for real-time AI-powered audio and video interaction with Huawei Cloud.

Soul brings its own speech synthesis and real-time video generation to the solution, building on its experience of emotional perception and multimodal interaction.

SoulX Builds the Foundation for Real-Time Interaction

SoulX is Soul’s voice-first, emotion-aware model for multimodal interaction. It’s all about emotional perception and voice and is continually improved as it’s used in the real world on the Soul app.

SoulX combines speech understanding, emotional expression, dialogue management, and multimodal generation. It has speech synthesis, singing voice synthesis, voice conversion, multi-speaker transcription, real-time turn-taking and dialogue control, real-time avatar generation, and human animation. The aim is to enable AI products to have conversations, show emotion, and interact in a more natural way.

Soul has open-sourced a number of their SoulX sub-models for voice, avatar and interaction tasks. In October 2025, they launched SoulX-Podcast, a model for podcast-style speech. The day after launch, it was at the top of Hugging Face’s TTS trending list. Their GitHub page has now amassed over 3,500 stars.

Later on, Soul released SoulX-FlashTalk for audio-driven avatars, SoulX-FlashHead for streaming talking heads, and SoulX-LiveAct for real-time human animation. SoulX-Singer produces singing voices, SoulX-Duplug deals with dialogue and turn-taking, and SoulX-Transcriber is the multi-speaker transcription tool. Soul is continuously improving the suite of tools and working on practical solutions for businesses.

Bringing AI Interaction Into Everyday Use

This is a solution that combines Soul’s real-time interaction technology with Huawei Cloud’s infrastructure to deliver more natural and responsive voice and video interactions. Huawei Cloud provides the requisite cloud and data infrastructure, ModelArts for training and inference, and high-performance storage.

On Soul App, SoulX enables AI-powered expression and social features to help users find relationships, initiate conversations, and build connections. Soul is also extending these technologies to portable smart devices. At HUAWEI CONNECT 2026, they demonstrated how these features can work on everyday hardware.

Portable smart devices make it very easy for people to talk to AI via voice at any time they wish. Soul is also looking at ways to make multimodal interaction on these devices more proactive and to maintain the conversation over time.

From Product Experience to Industry Solutions

At the roundtable on September 18th at HUAWEI CONNECT 2026, Soul spoke about its thoughts on AI hardware and how emotion-aware models can influence the user experience.

Since launching Soul App in 2016, we’ve learned a lot about how people express themselves, how they build relationships, and how they interact. That experience has been the basis for the development of SoulX and the various other solutions. Those features are now being rolled out to smart devices so that voice-based AI interaction becomes part of everyday life.

For industry partners, Soul provides real-time interaction technology with the benefit of having actually built and shipped real-world products. You can add voice interaction and digital avatars to smart hardware, embodied AI, and media/creative applications.

Soul also has governance in its models and applications. The safety framework covers digital avatars, voice interaction, and generated content with features such as content safety alignment, AI labeling, minor protection, and anti-fraud. The partnership leverages Soul’s experience and Huawei Cloud’s infrastructure to enable partners to build AI products with more natural interaction and richer emotional expression.