Unisound Showcases IoT-Oriented AI Chip UniOne and Full-Stack Solutions at 2019 World Artificial Intelligence Conference

From August 29 to 31, the 2019 World Artificial Intelligence Conference was grandly held at the Shanghai World Expo Exhibition and Convention Center. Themed “Intelligent Connectivity for an Infinite Future,” the conference brought together nearly 300 leading enterprises in the AI sector, conveying the most cutting-edge technological developments to the world and showcasing China’s practical achievements in artificial intelligence.

As one of the few leading AI enterprises in China boasting a full-stack technological chain encompassing algorithms, computing power, and chip capabilities, Unisound traveled to Shanghai at the invitation of the organizers to participate in this premier global artificial intelligence conference. The company comprehensively showcased its “AI+” practical applications and core AI technologies.

AI chips are the heart of artificial intelligence. In the era of the Internet of Things (IoT), terminal devices are being endowed with increasing AI capabilities. To activate the intelligent connectivity of everything while performing AI computations under constraints of low power consumption and low cost, a powerful hardware foundation is essential. To this end, Unisound has spent three years overcoming traditional barriers through intensive technological R&D, independently developing UniOne, an AI chip designed for IoT, along with its “Swift” solution.

This chip features an ARM core and a proprietary CPU+uDSP+DeepNet architecture, supporting 8/16-bit vector and matrix operations. Built on a deep learning network architecture, it maximizes parallel computing performance for voice AI, meeting the demands of massive terminal devices in terms of cost, stability, and integration. It provides artificial intelligence technologies such as speech recognition, natural language understanding, and speech synthesis for IoT devices like smart home systems, children’s intelligent companion robots, and smart speakers, increasing computing power by more than 50 times while significantly reducing costs and power consumption.

Nowadays, the integration of AI into the Internet of Things (IoT) has become an irresistible trend. In an era defined by the principle “No Chip, No AI,” it is no easy feat for either upstream hardware manufacturers or downstream technology service providers to fully embrace AI. Leveraging its Swallow platform, Unisound offers not only high-performance chips and terminal engines but also turnkey solutions tailored to diverse user scenarios. This approach enables clients to start from a higher design baseline, open-source and optimize full-stack voice interaction technologies built upon the chip, thereby empowering a vast number of IoT device manufacturers and solution providers to build a richer AIoT ecosystem.

As a leader in the implementation of medical voice interaction solutions, Unisound has completed the strategic upgrade of its medical product line from perception to cognition this year, advancing from recognition to understanding and decision-making. By targeting different stages and scenarios of clinical diagnosis and treatment, it aims to invigorate medical services and drive the intelligent transformation of the healthcare industry.

On the day of the conference, Unisound prominently showcased its intelligent waiting-room robots, medical voice interaction solutions, and intelligent medical record quality control systems, which cover multiple stages including pre-consultation, during-consultation, and post-consultation.

Intelligent waiting room robots can proactively initiate and guide conversations with patients prior to consultations, completing medical history collection and sending condition summaries (medical histories) to physicians. This helps doctors understand patients’ conditions in advance, thereby improving consultation efficiency and reducing misdiagnosis. During consultations, the medical voice interaction solution assists physicians with voice-based case entry, enhancing documentation efficiency. Built on AI technologies such as deep learning, high-performance computing, and big data, this solution employs an intelligent medical language model that enables the speech recognition engine to achieve fast and accurate recognition, robustness against accents, and strong noise immunity. For post-consultation care, the intelligent medical record quality control system accurately interprets the content of medical records and screens for deficiencies, reengineering business processes to significantly improve the efficiency, depth, and breadth of quality control, thereby ensuring the quality of medical services and patient safety.

Currently, Unisound’s smart healthcare-related products have been deployed in over 100 hospitals across China, with testing completed in more than 500 additional hospitals.

This year, Unisound has officially partnered with Shimao Group, committing to the deep integration of smart technology with the real estate industry. Guided by a user-centric approach, they aim to create a high-quality smart living experience that spans from smart homes to intelligent communities.

At the conference, to allow attendees to tangibly experience the capabilities of the SwiftChip and the technological sophistication of smart home solutions, Unisound set up a 1:1 physical model home on-site. Equipped with the “Xiao Mao” smart speaker jointly developed by Unisound and Shimao Group, the showcase enabled voice-interactive control and adjustment of in-room devices, offering a firsthand glimpse into whole-home intelligent connectivity.

“Xiao Mao” smart speaker is an AI voice butler for guest rooms, built on the Swift chip. Powered by Unisound’s advanced artificial intelligence technology, “Xiao Mao” achieves ultra-long-range voice response, allowing easy activation from any corner of the room. It supports streaming interaction, intelligently understands contextual cues, and allows conversation interruption. A single wake-up command initiates multi-turn dialogues. With OTA (Over-the-Air) online updates, no manual intervention is required. Equipped with over 300 smart skills, it serves as an intelligent butler that fulfills both “guest control” and “guest service” needs.

Unisound has remained deeply committed to the “AI + Education” sector, providing robust technical solutions and comprehensive service support for the children’s intelligence domain. This time, Unisound’s booth welcomed a new partner—the Congcong AI Companion Education Robot. Offering functionalities such as listening, speaking, viewing, companionship, reminders, and learning assistance, this robot marks Unisound’s first AI hardware product targeted at the consumer market since its founding seven years ago. It also serves as a flagship example of the deployment of Unisound’s KEROS 2.0 Intelligent Education Operating System onto hardware devices.

Congcong features a 7-inch HD blue-light-filtering touchscreen, a T-shaped 4-microphone array supporting 360° sound source localization and 5-meter far-field voice pickup. The device is equipped with a large-aperture wide-angle camera, ambient light sensor, proximity sensor, gravity sensor, and a 350° high-precision servo motor. It also has a built-in 4400mAh battery, enabling cordless operation.

At the event, Congcong’s adorable and endearing appearance attracted many attendees to stop and interact. In standby or conversation mode, its lively large eyes on the screen can display various expressions such as hearts, frowns, and anger. It also supports sound source localization and face-tracking features, keeping its gaze fixed on the user at all times to enhance emotional engagement with children. In terms of AI-powered educational functions, Congcong not only performs image-based environmental recognition and analysis but also offers features such as AI-assisted reading, spoken language assessment, phonological awareness启蒙 (early literacy), idiom solitaire, AI fruit recognition, and AI translation. Through multimodal interaction, it effectively boosts children’s interest in learning.

Shaping the Future with Core Innovation. Unisound will continue to deepen its engagement in user scenarios, explore multi-domain “AI+” models, and empower industrial commercialization with full-stack, hard-core AI technologies. The 2019 Artificial Intelligence Conference is now in full swing. You are welcome to visit the Unisound booth (Hall 2, Booth C188) at the Shanghai World Expo Exhibition & Convention Center to experience smart technology.