We’ve spent years collecting & annotating speech data. What became equally important was learning how to evaluate speech models in a way that reflects real listeners, real accents, & real usage. This video explains how we ran this blind evaluation & what @JoshTalksLive is building in speech evaluations.

@JoshTalksLive So much potential here, communication between humans can radically change over next few years

@JoshTalksLive blind evals kill the lab-to-real dropoff. accents and messy usage introduce entropy that exposes weak scaling laws, just like unchecked drift in production pipelines. this is listener sovereignty verified. 🇳🇴