Massive Sound Embedding Benchmark (MSEB)

August 21, 2026 ยท View on GitHub

Unittests PyPI version

This is not an officially supported Google product.

The Massive Sound Embedding Benchmark (MSEB) provides a framework for assessing sound embedding methods across diverse sound categories and tasks that leverage improved sound representations.

The published leaderboard can be found at: https://huggingface.co/spaces/google/mseb-leaderboard