Category Research

Congratulations to AIM members Aditya Bhattacharjee and Christos Plachouras, and C4DM member Sungkyun Chang who secured first place at the Query-by-Vocal Imitation (QbVI) Challenge, held as part of the AES International Conference on Artificial Intelligence and Machine Learning for Audio conference (AES AIMLA 2025) taking place from September 8-10, 2025.

The winning entry addressed the task, which entails retrieving relevant audio clips from a database using only a vocal imitation as a query. This is a particularly complex problem due to the variability in how people vocalise sounds and the acoustic diversity across sound categories. Successful approaches must bridge the gap between vocal and non-vocal audio, while handling the unpredictability of human-generated imitations.

The team’s submission, titled “Effective Finetuning Methods for Query-by Vocal Imitation”, advances the state-of-the-art in QbVI by integrating a triplet-based regularisation objective with supervised contrastive learning. This method addresses the issue of limited data by sampling from an unused subset of the VocalSketch dataset, which comprises practice recordings and human-rejected vocal imitations. While this data may not be suitable for positive matches, the vocal imitation data is useful for creating confounding examples during training. More specifically, this increases the size of the pool of negative examples, which is utilised by the added regularisation method.

The proposed method surpassed state-of-the-art methods for both subjective and objective evaluation metrics, opening up scope for product-based innovations and software tools that can be used by artists to effectively search large repositories of sound effects.

On 21-25 September 2025, several AIM researchers will participate at the 26th International Society for Music Information Retrieval Conference (ISMIR 2025). ISMIR is the leading conference in the field of music informatics, and is currently the top-cited publication for Music & Musicology (source: Google Scholar). This year ISMIR will take place onsite in Daejeon, Korea.

Similar to previous years, AIM will have a strong presence at ISMIR 2025.

In the Scientific Programme, the following papers are authored/co-authored by AIM members:

• Audio Synthesizer Inversion in Symmetric Parameter Spaces with Approximately Equivariant Flow Matching (Ben Hayes, Charalampos Saitis, György Fazekas)
• SLAP: Siamese Language Audio Pretraining without Negative Samples for Music Understanding (Julien Guinot, Alain Riou, Elio Quinton, György Fazekas)
• GD-Retriever: Controllable Generative Text Music Retrieval with Diffusion Models (Julien Guinot, Elio Quinton, György Fazekas)
• Instruct-MusicGen: Unlocking Text to Music Editing for Music Language Models via Instruction Tuning (Yixiao Zhang, Yukara Ikemiya, Woosung Choi, Naoki Murata, Marco A. Martínez Ramírez, Liwei Lin, Gus Xia, Wei Hsiang Liao, Yuki Mitsufuji, Simon Dixon)
• Scaling Self Supervised Representation Learning for Symbolic Piano Performance (Louis Bradshaw, Honglu Fan, Alexander Spangher, Stella Biderman, Simon Colton)
• Codicodec: Unifying Continuous and Discrete Compressed Representations of Audio (Marco Pasini, Stefan Lattner, György Fazekas)
• MIDI-VALLE: Improving Expressive Piano Performance Synthesis through Neural Codec Language Modelling (Jingjing Tang, Xin Wang, Zhe Zhang, Junichi Yamagishi, Geraint Wiggins, György Fazekas)
• Perceptual Errors in Music Source Separation: Looking Beyond SDR Averages (Saurjya Sarkar, Victoria Moomjian, Basil Woods, Emmanouil Benetos, Mark Sandler)
• GOAT: a Large Dataset of Paired Guitar Audio Recordings and Tablatures (Jackson Loth, Pedro Sarmento, Saurjya Sarkar, Zixun Guo, Mathieu Barthet, Mark Sandler)
• CMI-Bench: a Comprehensive Benchmark for Evaluating Music Instruction Following (Yinghao Ma, Siyou Li, Juntao Yu, Emmanouil Benetos, Akira Maezawa)
• Assessing the Alignment of Audio Representations with Timbre Similarity Ratings (Haokun Tian, Stefan Lattner, Charalampos Saitis)
• Improving Neural Pitch Estimation with SWIPE Kernels (David Marttila, Joshua D. Reiss)
• Refining Music Sample Identification with a Self Supervised Graph Neural Network (Aditya Bhattacharjee, Ivan Meresman Higgs, Mark Sandler, Emmanouil Benetos)

The following Tutorials will be co-presented by AIM PhD students Rodrigo Diaz and Julien Guinot:

• Differentiable Physical Modeling Sound Synthesis: Theory, Musical Application, and Programming (Jin Woo Lee, Stefan Bilbao, Rodrigo Diaz)
• Self-supervised Learning for Music – An Overview and New Horizons (Julien Guinot, Alain Riou, Yuexuan Kong, Marco Pasini, Gabriel Meseguer-Brocal, Stefan Lattner)

The following journal papers published at TISMIR which are co-authored by AIM members will be presented at the conference:

• Predicting Eurovision Song Contest Results: A Hit Song Science Approach (Katarzyna Adamska, Joshua Reiss)
• The GigaMIDI Dataset with Features for Expressive Music Performance Detection (Keon Ju Lee, Jeff Ens, Sara Adkins, Pedro Sarmento, Mathieu Barthet, Philippe Pasquier)

As part of the MIREX public evaluations:

• AIM PhD student Yinghao Ma is task captain for the Music Reasoning QA, Audio Beat Tracking, and Audio Key Detection tasks
• AIM PhD student Huan Zhang is task captain for RenCon 2025: Expressive Performance Rendering Competitionk

Finally, on the organisational side:

• AIM PhD student Chin-Yun Yu is Virtual Co-Chair for the ISMIR 2025 conference.
• AIM PhD student Yinghao Ma is co-organising the satellite workshop LLM4MA: Large Language Models for Music & Audio

See you at Daejeon!