Have you tested their new Chirp v2 model? Curious if there's any improvement there.

>the only hyperscaler with a good ASR is Azure

How would you say the non-hyperscalers compare? Speechmatics for example?