SHANGHAI, May 28, 2026 (GLOBE NEWSWIRE) -- FinVolution Group officially launched the 2026 FinVolution Global Data Science Competition in May 2026. This year's challenge focused on turn-taking modeling in conversations, with the aim of giving voice AI a sense of when to speak.
Voice interaction has reached the native-audio era, with AI now responding in real time. What it still lacks is something humans do by instinct: knowing when to take a turn, when to stay silent, and when a brief "mm-hm" is the right reply. Without it, even the fastest model talks over users or lets dialogue stall.
This year's participants will be given thirty seconds of dual-channel dialogue as context, predict the speech events likely to occur in the next 800 milliseconds, equipping AI with the social intuition to read user intent and respond at the right moment.
The dataset behind the challenge is built from real dual-channel telephone conversations recorded across 35 regions of China, spanning a wide range of dialects and speaking styles. Audio comes paired with ASR transcripts and word-level timestamps, allowing participants to build pure-audio or multimodal systems.
"For more than a decade, this competition has been our way of connecting academic research with real-world application," said Tiezheng Li, CEO of FinVolution Group. "Turn-taking is one of the open problems in voice interaction today. We hope what's built here reaches far beyond research, letting millions of users experience more natural, more human conversation in everyday life."
The 2026 challenge is supported by the China Computer Federation(CCF) Technical Committee on Natural Language Processing as academic advisor, in collaboration with Fudan University's Natural Language Processing Lab. It is also an official partner competition of the 15th CCF International Conference on Natural Language Processing and Chinese Computing(NLPCC 2026). Top-performing teams will earn a direct path to present at NLPCC 2026 alongside the global NLP research community.
Competition Timeline
The 2026 challenge offers a prize pool of RMB 308,000 (approximately USD 42,900) and will unfold in three stages:
Preliminary Round (May 13 – June 19): Participants train locally and submit their prediction results for real-time scoring by the end of June 16. The list of teams advancing to the semifinals will be announced on June 19.
Semifinals (June 20 – July 16): Semifinalists are required to submit Docker images for evaluation by the end of July 7. Finalists will be announced on July 16.
Final Round (July 16 – late July): Finalists will compete in person before a panel of judges, with the exact date to be announced separately.
Registration is now open through the official competition platform：https://ai.ppdai.com/mirror/show?channel=media2
Over its eleven editions, the FinVolution Global Data Science Competition has drawn close to 10,000 participants from universities, research labs, and technology companies around the world. Past challenges have spanned deepfake detection, credit scoring, fraud detection, user behavior modeling, and dialect recognition. Partnerships with top AI conferences such as IJCAI (2024), CIKM (2025), and now NLPCC (2026) reflect its rising stature in the global AI ecosystem.
About FinVolution Group
FinVolution Group (NYSE: FINV) is a leading fintech company connecting millions of consumers and small businesses with financial institutions through advanced credit technology. Founded in 2007 and listed on the New York Stock Exchange in 2017, the Company operates across China, Indonesia, the Philippines, Pakistan, and Australia, with longstanding work in AI, big data, fraud detection, and credit risk modeling. FinVolution actively supports academic research through long-running sponsorships of premier AI conferences including WWW, IJCAI, CIKM, and NLPCC.
FinVolution Group
Zhou Zihui
Email: [email protected]
Website: https://www.xinye.com/
Telephone: +86-17764188164
City: Shanghai
A photo accompanying this announcement is available at https://www.globenewswire.com/NewsRoom/AttachmentNg/13f09984-5ace-47e3-a78b-5f57568637b9