I do not want my audio leaving my computer
Why it happens
The cloud engines are the fastest, so they are used for the other side's voice by default. Local Privacy transcribes both sides on your Mac, and transcription then uses no credits.
How to fix it
Open Settings → Audio & Models → Speech Recognition and choose Local Privacy.
Local Privacy runs the Local Whisper Model on your Mac, and you can trade accuracy against memory with the model size.
Answers still need the cloud: transcript text, never audio, is sent to the answer model.
