Comment by adamcharnock

FWIW - I also really like Wispr Flow, but I moved to running the 'Whisper Large' model locally using Handy (https://github.com/cjpais/Handy), which has been essentially as good, while also having lower latency.