Hi everyone,
I’ve been using the built-in Mac dictation for a while, but I’m finding it a bit too slow and not always accurate. So I’m on the lookout for something that either lets you plug in a corporate ChatGPT license or just does a better job on-device. Any recommendations I should check out?
I have had great success with an app called MurmurType. You can give it a try with Setapp if you use that at all, or you can check out their page. I’ve used it quite often, and find it to be quite accurate even with my accent.
I’ve been using Superwhisper for a few months now, and I find it to be excellent. You can choose your preferred AI model, so perhaps that would align with your corporate ChatGPT license.
Superwhisper is the best I’ve tried. It’s a subscription, and requires continual training for vocabulary and word spellings (as they all do). But it has many options for voice recognition and AI processing models based on speed and accuracy, includes the AI models in its subscription fee, and still works very well out of the box for general use.
Agree that it works very well out of the box. FYI: There’s also an option for a lifetime purchase if you don’t want to pay for a subscription.
Superwhisper and Spokenly both let you bring your own API key, but that is via the API, not using a regular account login, such as ChatGPT Plus.
Now that I’ve started using SuperWhisper after migrating from VoiceInk, I’m curious about your setup. Could you share the different modes you use, along with the prompts for each and the models you’ve selected? Any insights would be greatly appreciated.
I actually just use the ‘default’ mode, with no particular prompt in SuperWhisper. When I first set it up I started creating different prompts (“modes”) for different contexts, but I found I preferred the tool to simply capture my words as they were spoken, and then edit the text myself. However, at some point I intend to create a mode with instructions for capturing punctuation and line breaks. I appreciate that these modes could probably save me a step/time, but for now I’m happy to just capture what I have said.
The one thing I’ve adjusted is to continually add vocabulary for words/phrases that I use that it misunderstands.
As to which models I use, I don’t have any strong preference, but use “Ultra” as the voice model.
Gemini, groq, cerebras offer pretty good free tier. Gemini is the most liberal. No rate limits whatsoever lol.
Gemini 3 flash is an absolute beast and super underrated. Tt beats Claude sonnet 4.5 and haiku for me.
Why did you migrate from Voiceink to Superwhisper ?