All aboard the voice to text train - The Desk of Bryan Maniotakis

"TypeWhisper is an open-source, native Mac app, with a ton of options that should support whatever your preferred LLM provider is. It has a clean macOS interface with a floating live transcription bar at the bottom of the screen, which is helpful for seeing what it's picking up in real time (though it's optional). You can set a global hotkey to automatically trigger the TypeWhisper input. I've set mine using a hyperkey mapped to Caps Lock + Spacebar, so from any app I can press that combination and it will automatically start transcribing."

"Once transcription is complete, it runs through Claude with specific rules to clean up filler words and fix grammar while keeping the core sentence structure intact. The result is clean text ready to paste directly from my clipboard. The full interaction is just Caps Lock + Spacebar, then paste. Done."

"It also has a few other great features, like file transcription. I can drop audio or video files directly into the app and it will automatically extract the text. You can also have it monitor selected folders, so if I want to add a class audio recording, it can automatically transcribe that for me as well. It supports multiple languages, markdown output, and much more."

"One standout feature that many competing apps lack is the ability to add words to a custom dictionary. Since I work in the tech space, a lot of the terminology I use won't appear in a standard dictionary. TypeWhisper ships with pre-installed vocabulary libraries that you can toggle on and off - web development, iOS, macOS, data and AI, Figma design, and more are"

TypeWhisper is an open-source native Mac voice-to-text app with a macOS interface that can show an optional floating live transcription bar. A global hotkey can trigger transcription from any app, and after transcription finishes it runs the text through Claude with rules that remove filler words and correct grammar while preserving the core sentence structure. The cleaned output is ready to paste directly. The app also supports transcribing audio or video files by dropping them into the app, and it can monitor selected folders for automatic transcription. It supports multiple languages and markdown output. It includes custom dictionary support with toggleable vocabulary libraries for domains like web development, iOS, macOS, data and AI, and Figma design.

Read at The Desk of Bryan Maniotakis