Artificial
Musical
Intelligence

An open platform created to advance research on artificial musical intelligence. Explore models, benchmarks, and more.

View on GitHub

Benchmarks

Commonly used benchmarks.

Click to see more details. Press Select, then pick multiple to cite or download in bulk. Benchmarks with questions loaded on the site can be inspected.

← All benchmarks

Questions

Models

An ever-growing list of Audio-Language Models (ALMs) or multi-modal models that listen to audio.

Click a model for release date, creators, and paper.

Evaluation

Click a category to see its full definition, coverage, example, and evaluation methods.

Evaluation study

The MMAR music set, answered by two models both as multiple choice (apparent skill) and open-ended (actual skill), with open-ended answers graded by a PIAC-aware judge.

By PIAC category

Results

Published benchmark performance, linked at the level of each reported value to its original source.