1794571413-logo-v2
SIGN IN

Highlights from the Data for Voice AI Meetup in Zhuhai

1750128659-英文logo带背景

Posted at 14 hours ago

On the evening of November 5, we gathered with friends who share our passion for speech, language, and open source to celebrate MagicHub.com's fifth anniversary.

The Meetup provided an opportunity for community members to connect and exchange ideas. Some shared their ongoing research and sought feedback from peers, others talked about their new plans, and some met people with similar research interests along the way.

Here are the standout moments from our Zhuhai stop 👇

1. Lightning Talk Highlights: Fresh Ideas and Technical Insights

  • Defining "Wo Qu" (我去) from a Pragmatics Perspective: One of the most engaging talks came from Wang Zepeng, a postdoctoral researcher at Wuhan University, who analyzed the Chinese expression "Wo Qu" (我去)—a phrase similar to "Holy cow!" or "No way!"—through the lens of pragmatics. Depending on the context, the expression can convey surprise, serve as a way of expressing frustration, or function as a discourse marker. Even an everyday expression like this involves complex linguistic mechanisms, including the speaker's intention, the listener's interpretation, and the context in which it is used. It also raises a question: how can AI understand the "implicit meaning" behind human language?

  • Multilingual Research in the Mekong Region: Liu Changyuan from Tongxin Zhiyu Technology shared insights into multilingual research in the Mekong region. By discussing the challenges involved in data collection, annotation, and model training across different languages, the talk introduced attendees to research questions worth exploring beyond their own familiar areas of work.

  • Physical Intelligence Also Belongs to Speech Interaction: Zhou Lang from ArkAI shared his exploration of "bringing physical intelligence to the world around us". As intelligent systems enter the physical world, how sound travels and how machines communicate with people also need to be carefully considered.

  • Voice Agents in Practice: Gu Shuhua, a recent graduate of the University of Macau, shared his plans to start a One-Person Company (OPC) focused on Voice Agents. Sharing a new project with peers in person and hearing their feedback is part of what makes gatherings like this meaningful.

Each of the four talks offered a different perspective and sparked lively conversations around the topics that interested attendees most. That was what made the dinner meaningful: it gave people who usually connect online a chance to meet in person and get to know one another.

For anyone who missed out this time, there is no need to worry—NeurIPS heads to Sydney, Australia this December, so stay tuned for our upcoming Meetup there!

02. At NCMMSC 2026: Conversations About Data at Our Booth

From November 6 to 8, Magic Data and MagicHub.com also set up a booth at NCMMSC 2026.

Many faculty members, students, and industry partners stopped by to learn about our open-source community and data services, discussing data-related challenges in their research. Based on their specific needs, we explored dataset selection and possible solutions together, and introduced opportunities to get involved in the open-source community to those who were interested.

For the MagicHub.com open-source community, these face-to-face conversations were particularly valuable. Learning about the research people are working on, their data needs, and the support they are looking for offers meaningful insights into how we can continue building the community.

03. Student Forum: Stay Passionate, Embrace What's Next

At the NCMMSC 2026 Student Forum, Dr. Zhang Qingqing, CEO of Magic Data, addressed a question many students are thinking about: What kind of people will companies need in the future?

As technology continues to advance, the job scopes are changing as well. Once you join a company, your responsibilities may go beyond what is listed in your job description. Confronting unfamiliar problems, these capabilities—understanding new situations actively, finding workable solutions, and collaborating across teams—are becoming increasingly important.

This also means that personal growth needs to be considered in the real work: understanding your strengths and how you can contribute to the team, as well as being willing to learn, try new things, and take on new responsibilities when faced with unfamiliar challenges. These capabilities are built through ongoing collaboration and problem-solving.

Dr. Zhang's talk also touched on three qualities, "passion, empathy, and creativity". Those are the exact qualities we look for in teammates: diving deep into their field, understanding users and teammates, and stepping forward with fresh ideas when new challenges arise.

END | Closing: Until We Meet Again

The MagicHub.com 5th Anniversary Meetup in Zhuhai has come to a close, and we would like to thank everyone who joined us and made it special. Gathering in Hengqin to share research and projects, discuss data needs, and get to know one another in person turned many online connections into face-to-face encounters.

Over the past five years, MagicHub.com has continued to grow with the participation and support of the community. We will continue making datasets openly available and look forward to hosting more Meetups where community members can connect in person, exchange ideas, and find like-minded collaborators.Our Meetup in Zhuhai left us with many topics worth continuing to explore. We look forward to meeting again at our next stop, the Full-Duplex Voice AI Meetup @ NeurIPS in Sydney, where we can hear about everyone's latest progress and meet more new friends.

Related Datasets

Datasets Download Rank

ASR-RAMC-BigCCSC: A Chinese Conversational Speech Corpus
Multi-Modal Driver Behaviors Dataset for DMS
ASR-SCCantDuSC: A Scripted Chinese Cantonese (Canton) Daily-use Speech Corpus
ASR-CCantCSC: A Chinese Cantonese (Canton) Conversational Speech Corpus
ASR-SCSichDiaDuSC: A Scripted Chinese Sichuan Dialect Daily-use Speech Corpus
ASR-SCCantCabSC: A Scripted Chinese Cantonese (Canton) Cabin Speech Corpus
ASR-CShhiDiaCSC: A Chinese Shanghai Dialect Conversational Speech Corpus
ASR-EgArbCSC: An Egyptian Arabic Conversational Speech Corpus
ASR-CSichDiaCSC: A Chinese Sichuan Dialect Conversational Speech Corpus