WIMIR virtual meetup
We are happy to announce the next WIMIR virtual meetup, happening on Saturday, June 13, 2026!
The meetup is meant for WIMIR mentees and mentors to connect in a casual setting, share about their work and/or personal experiences, and hopefully engage in some enriching conversations. This year, we opened the meetup to both the WIMIR and ISMIR communities, with the aim of exposing our mentees and mentors to a wider audience and broadening the discussions with new perspectives.
Date: Saturday, June 13, 2026
Time:
Two sessions, 2 hours each. You are free to join either or both sessions, or join and leave at any time.
Session 1: 8:00-10:00 UTC, starts at:
- 9:00: UK (BST)
- 10:00: Germany/Spain (CEST)
- 11:00: Turkey (TRT)
- 13:30: India (IST)
- 16:00: China (UTC+8)
- 18:00: Sydney, Australia (AEST)
📅 [Download ics] [Google Calendar]
Session 2: 16:00-18:00 UTC, starts at:
- 9:00: CA, USA (PDT)
- 10:00: Mexico (CST)
- 11:00: IL, USA (CDT)
- 18:00: Germany/Spain (CEST)
📅 [Download ics] [Google Calendar]
Zoom link: To be announced one day before the event (stay tuned!)
No registration is needed. Feel free to join anonymously if you feel more comfortable doing so. Presentations will not be recorded.
For any questions or comments, contact us at wimir-mentoring@ismir.net
Program:
| ⏰Session 1 | |
| 00:00-00:10 | Welcome |
| 00:10-00:25 | T1. MusExam: A Browser-Native System for Constraint-Driven Music Notation and Interactive Theory Assessment Hale Fulya Celikel Soganci, Musicologist, Lidyanka, Istanbul, Turkey |
| 00:25-00:40 | T2. Rethinking Evaluation of Self-Supervised MIR Models Arjun Bahuguna, Master’s student, Universitat Pompeu Fabra (UPF), Barcelona, Spain |
| 00:40-00:55 | T3. Time on Earth Jordan Smith, Lecturer, Queen Mary University of London (QMUL), London, UK |
| 00:55-01:10 | T4. Beyond Papers: My Missing Curriculum Lele Liu, Postdoc, Julius Maximilians University of Würzburg (JMU), Würzburg, Germany |
| 01:10-02:00 | Discussions |
| ⏰Session 2 | |
| 00:00-00:10 | Welcome |
| 00:10-00:25 | T5. Coding Without a Computer? Anna Guadarrama, Researcher, independent, Mexico City, Mexico |
| 00:25-00:40 | T6. Automatic Mixing, Fx-Representations, and the WIMIR mentee experience Vansh Chugh, Undergraduate student, Knox College, IL, USA – |
| 00:40-00:55 | T7. “I Have No Idea, But Let’s Do It Anyway”: Breaking the Beast of Agentic AI Mahya Khazaei, Machine learning researcher, Water&Music, Toronto, Canada – |
| 00:55-01:10 | T8. I really wanted to do a PhD Rhythm Jain, Scientist, SiriusXM, CA, USA |
| 01:10-02:00 | Discussions |
Abstracts:
T2. Rethinking Evaluation of Self-Supervised MIR Models
Arjun Bahuguna, Master’s student, Universitat Pompeu Fabra (UPF), Barcelona, Spain
Standard evaluation metrics for MIR tell you whether a model works, not why it works. For SSL models this is especially consequential: downstream accuracy conflates representation quality with probe capacity, meaning a model can score well on benchmarks while harbouring representational failures the metrics cannot see. To fix this, we show that embedding geometry and information-theoretic measures explain downstream performance of joint text-audio embedding models like CLAP better than standard benchmarks.
T3. Time on Earth
Jordan Smith, Lecturer, Queen Mary University of London (QMUL), London, UK
Thank you for your time! Mentorship involves giving other people time, which is precious, because time is a resource we can never replenish. Our time on Earth is bounded, and this is reflected poetically by any piece of music, which also only exists in time, and also must end if it ever begins. If you attend this talk, your time will be spent listening to me expand on this connection. If I’m quick, I’ll manage to squeeze in World War II, magic, the Fourier Transform, education, crystals, aliens, and — uh oh, it says here “A.I.”!
T4. Beyond Papers: My Missing Curriculum
Lele Liu, Postdoc, University of Würzburg (JMU), Würzburg, Germany
When I started my PhD, I thought research was mostly about conducting experiments and writing papers. Over time, I realised that many of the lessons that shape the development of a researcher came from somewhere else: communicating with supervisors and collaborators, participating in research communities, reviewing papers, making work visible and accessible, and gradually developing a research identity. In this short reflective talk, I will share a few lessons, surprises, and ongoing questions from my own journey as an early-career researcher. Rather than presenting answers, I hope to invite discussion across career stages and research sectors. What are the most important things we learn as researchers that were never part of the official curriculum? And what do we wish someone had told us earlier?
T5. Coding Without a Computer?
Anna Guadarrama, Researcher, independent, Mexico City, Mexico
A classical musician learns MIR: how does a person who has 20+ years of experience in music become a beginner again? Being used to practicing 8 hours of piano a day, I transferred this to sitting 12 hours a day at a computer during the summer vacation. I already had a bit of experience with python basics, but everything ended in error. I couldn’t tell what was right from wrong in my code. This led me to be radically honest with myself: I am no expert and I can’t do things alone. I started asking for more help from my WIMIR mentor, Dr. Román, and instead of being disappointed at my lack of expertise, he was excited and supportive. This allowed me to try out new things. One day, the internet went out and I started writing my code on a plain sheet of paper. Something clicked at that moment, writing out those meaningless symbols by hand—they came alive. Instead of the immediacy we get from writing out something digitally, this slowed me down to get to know each line as if it was its own musical motif with a hidden meaning. Being entry level for the first time since I was 6 years old made me realize how getting past the “ego” of being an expert can make a difference between feeling stressed and having fun—and having a program that runs!
T7. “I Have No Idea, But Let’s Do It Anyway”: Breaking the Beast of Agentic AI
Mahya Khazaei, machine learning researcher, Water&Music, Toronto, Canada
My MIR journey began right here in the WiMIR community. But even with a great network, the real world moves fast. While pouring my heart into singing voice conversion for my Master’s, the AI boom hit overnight. I’d open Instagram to see deepfake politicians flawlessly singing folk songs while my own models struggled, and honestly, I lost hope. How do you keep going when tech outpaces you? You embrace the mindset: “I have no idea, but let’s do it anyway.” In this talk, we’ll demystify agentic AI and break the beast of it so we aren’t afraid to use these tools. After a brief look at how agents are changing MIR, I’ll share the blueprint of a work-in-progress agent I’m building strictly for my own use. My biggest advice? Build for yourself first. Be your first consumer!
WIMIR blog
⚠️ This blog is under construction. Stay tuned for updated content!
💗 WIMIR has now grown from Women in MIR to Widening Inclusion in MIR. We welcome all underrepresented communities, including women, as well as trans and non-binary individuals, people from the Global South, and people with disabilities.
🤗 Want to be part of the WIMIR community? Join our group here: https://groups.google.com/g/wimir
🤝 Interested in the WIMIR mentoring program? Sign up here:
- Mentee form: https://forms.gle/ShvYwhTPqTvfoKUD9
- Mentor form: https://forms.gle/2nvW1adaYxUbVmsz5
💪 We are always looking for volunteers to help with various efforts in the group, including new mentors for the mentoring program, let us know if you would like to contribute: wimir-mentoring@ismir.net Let’s strenghten the MIR community together!
Widening Inclusion in Music Information Retrieval 