The Mac supports dictation. I thought it was all done locally, but that might not be the case for all things: “Dictation processes many English voice inputs on your Mac. Information will be sent to Apple in some cases, including for search.” This is according to the System Settings keyboard page.

MacWhisper Pro has been adding more and more audio capabilities, and I just noticed it now has a dictation setting. Once I gave it permission to access my system, I set it up to activate after pressing the right “option” key. This triggers the dictation and then does some processing with an AI service.

It turns out that you can add Ollama as an AI service to MacWhisper and choose a local model. I added “llama3.2:latest” as the Ollama service. Then back to the Dictation setting, I was able to select Ollama and when I tested dictation. Post-processing can happen specific to the current app you are using. Now all the dictation and any AI post-processing should be happening locally on my computer and I should be able to avoid the wishy-washy behavior of the built-in dictation.