Qfin Holdings' Intelligent Speech Team secured acceptance for their research paper at ASRU 2025, a premier global speech technology conference, highlighting the company's independent R&D capabilities and setting a new benchmark in multimodal emotion computing.

Qfin Holdings Achieves Milestone in Multimodal Emotion Computing, Paper Accepted by ASRU 2025

SHANGHAI – Qfin Holdings (QFIN) has marked a significant achievement in the field of artificial intelligence, with its Intelligent Speech Team securing acceptance for their research paper, "Qieemo: Multimodal Emotion Recognition Based on the ASR Backbone," by ASRU 2025, a flagship global conference in the speech technology industry.

Event in Detail

This recent development underscores Qfin Holdings' commitment to advanced research and development in AI. The acceptance by ASRU 2025 solidifies the company's standing among a select group of fintech pioneers recognized across the "Big Three" top-tier speech conferences, which include ICASSP, InterSpeech, and ASRU. This recognition positions Qfin Holdings within the global first echelon of speech technology research.

The research paper introduces a theoretical framework with universal significance for multimodal emotion recognition, moving beyond task-specific models. The technology developed by the Intelligent Speech Team demonstrates a substantial improvement in recognition accuracy, boosting it by over 15% compared to traditional methods. Furthermore, it achieved a 3.04% relative improvement over the existing state-of-the-art (SOTA) single-modal method, MSMSER, establishing a new "SOTA+" benchmark in emotion computing. This breakthrough enables intelligent customer service to possess genuine "emotional understanding" abilities for the first time.

Analysis of Market Reaction and Strategic Approach

The market's sentiment is potentially bullish for Qfin Holdings as this achievement signals strong R&D capabilities and potential for future competitive advantage in the rapidly evolving AI and fintech sectors. Unlike many fintech companies that rely on open-source solutions or external collaborations, Qfin Holdings adheres to an end-to-end independent R&D strategy in core artificial intelligence fields. This approach involves continuous investment in cutting-edge domains such as speech recognition and emotion computing, building a comprehensive system from algorithm design to engineering implementation.

This foundational research path, which delves into the mathematical principles and mechanisms of speech signal processing and multi-modal feature expression, distinguishes Qfin Holdings. While the industry often focuses on layering existing neural network architectures, Qfin Holdings' emphasis on basic theories and original frameworks provides significant advantages in technical depth, application flexibility, and long-term competitiveness. This strategy aligns with a broader market trend where investors are increasingly prioritizing companies with tangible AI infrastructure and proprietary technological advantages.

Broader Context and Implications

The success of Qfin Holdings' AI-Plus credit strategy, rolled out since early 2025, further illustrates the practical applications of its advanced AI research. This initiative aims to build the industry's first AI agent platform to empower core credit processes. The strategy has driven measurable improvements in operational efficiency, risk mitigation, and valuation multiples. In the first quarter of 2025, total loan facilitation volume grew by 15.8% year-over-year (YoY), reaching RMB88.88 billion, with a stable delinquency rate. Funding costs also declined, partly due to the AI-driven underwriting process that slashes approval times and costs.

While specific valuation ratios like Price-to-Earnings (P/E) or Price-to-Book (P/B) were not detailed in relation to this specific research achievement, Qfin Holdings' Second Quarter 2025 Financial Highlights demonstrate robust performance. Total net revenue reached RMB5,215.9 million (US$728.1 million), an increase from the prior quarter and the same period in 2024. Income from operations stood at RMB2,136.2 million (US$298.2 million), representing an operating margin of 41.0%. These figures, coupled with the strategic focus on proprietary AI, underscore the company's position as a compelling investment in China's fintech landscape.

"The company's breakthrough technology has boosted recognition accuracy by over 15% compared to traditional methods and achieved a 3.04% relative improvement over the existing SOTA single-modal method MSMSER, establishing a new 'SOTA+' benchmark in the field of emotion computing." - Industry Analysis

Looking Ahead

The acceptance of Qfin Holdings' research at a top-tier global conference and its ongoing investment in foundational AI R&D positions the company for continued innovation and competitive advantage. The ability of intelligent customer service to genuinely "understand emotions" through this technology could lead to the development of more sophisticated and user-centric financial products and services. As AI continues to reshape global finance, Qfin Holdings' emphasis on independent, end-to-end R&D suggests a strong trajectory for advanced product development, increased market share, and robust intellectual property in the long term. Future monitoring will focus on the practical implementation of these theoretical advancements into new offerings and their impact on market penetration and financial performance.