Voice to text with a push to talk workflow makes a lot of sense for developers who don't want constant background listening.
Curious how the local mode performs compared to the cloud version in terms of latency and accuracy