This repository contains the submission package of Query by Vocal Imitation challenge at the 2025 AES International Conference on Artificial Intelligence and Machine Learning for Audio.
Participants should clone this repository, and please ensure that you can successfully run AES-QVIM2025_submission_template.ipynb within the repository. Detailed instructions are provided in the template file.
For your final submission, please submit only the completed template file by email to qvim.aes@gmail.com using your registration email address, and rename the template as {teamName}_{submissionNumber}.ipynb. Please note that we only accept three submissions per team.
Your inference code will be run on Ubuntu (24.04) using a conda environment with Python 3.10. Please specify your conda environment if a different Python version is needed.
Additional packages must be installed with pip in the notebooks directly (see the two example notebooks).
You may use the functions provided in helpers.py; however, do not modify this file.
Prerequisites:
- Clone this repository.
git clone https://github.com/qvim-aes/qvim-submission.git
- Create and activate a conda environment with Python 3.10:
conda create -n qvim-submission python=3.10 jupyter
conda activate qvim-submission
python -m jupyterlab
Running the example notebooks should give the following results:
| Model Name | MRR (exact match) | NDCG (category match) |
|---|---|---|
| random | 0.0444 | ~0.337 |
| 2DFT | 0.1262 | 0.4793 |
| MN baseline | 0.2726 | 0.6463 |