Share your thoughts, 1 month free Claude Pro on usSee more

Home/Benchmarks

Speaker Separation on VoxCeleb2 (held-out)

97.8PIT Accuracy (MSE)

Echo

Updated 1d ago

Evaluation Results

Method	Links
Echo 2026.06		97.8	99.6	9.52

SOTA Paper

Echo

Echo: A Joint-Embedding Predictive Architecture for Speaker Diarization and Speech Recognition in a Shared Latent Space

Dataset

VoxCeleb

Follow for update

@wizwand_team Discord

Related Benchmarks

Speaker Diarization on VoxCeleb2 (held-out)Speaker Diarization on Simulated 2-spk mixtures Speaker Diarization on CALLHOME 2-spk Speaker Diarization on AMI IHM Speaker Diarization on CALLHOME part 2

© 2026 wizwand

Blog Contact Changelog Swarm

Privacy Policy Terms of Service FAQs Swarm Docs