Category Research

We are delighted to share that AIM PhD student Ben Hayes, along with AIM supervisors Charalampos Saitis and George Fazekas, have received the best student paper award at the ISMIR 2025 conference. The paper “Audio Synthesizer Inversion in Symmetric Parameter Spaces With Approximately Equivariant Flow Matching” proposes using permutation equivariant continuous normalizing flows to handle the ill-posed problem of audio synthesizer inversion, where multiple parameter configurations can produce identical sounds due to intrinsic symmetries in synthesizer design. By explicitly modeling these symmetries, particularly permutation invariance across repeated components like oscillators and filters, the method outperforms both regression-based approaches and symmetry-naive generative models on both synthetic tasks and a real-world synthesizer (Surge XT).

We are also happy to share that two AIM PhD students, Yannis Vasilakis and Ben Hayes, were recognised as outstanding reviewers.

This wraps up a fantastic week at ISMIR 2025 in Daejeon, South Korea, with very strong AIM participation. Pictured are current and past C4DM members, including many AIM members.

We are delighted to share that Mary Pilataki, a PhD student at AIM, has received the Best Paper Award at the Audio Engineering Society International Conference on Artificial Intelligence and Machine Learning for Audio (AES AIMLA) 2025.

The awarded paper, “Extraction and Neural Synthesis of Low-Order Spectral Components for Head-Related Transfer Functions”, presents research carried out during her internship at PlayStation London in collaboration with Cal Armstrong and Chris Buchanan. The study explores how deep learning can separate sound colouration from spatial cues in Head-Related Transfer Functions (HRTFs), opening up new possibilities for creating more personalised and immersive 3D audio experiences.

The full paper is available through the AES E-Library.

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.