This week we launched the Neurometric Audio Leaderboard.
https://leaderboard.neurometric.ai/
The leaderboard is designed to benchmark real-world performance of voice and audio Al models, with a focus on accuracy, efficiency, and practical deployment characteristics.
Our first update includes Modulate Velma 2, released yesterday. Velma is built using an ensemble-based approach that reflects a broader industry shift toward specialized, orchestrated systems.
We'll continue adding new models as they launch and sharing observations on how audio Al performance evolves over time.
