Speechcap is a Mac dictation application that operates entirely on the user’s computer, with an optional cloud transcription mode. Its primary function is to capture spoken input via a push-to-talk hotkey and convert it into formatted text that is inserted directly into the active application. The app uses OpenAI’s Whisper model for speech recognition and offers an on-device AI cleanup feature that removes filler words, restores punctuation, and applies context-aware formatting. Speechcap supports 89 languages for both transcription and translation. A free tier provides 2,500 words per week of raw transcription, while a Pro subscription (starting at $3 per month) unlocks unlimited dictation, AI cleanup, transforms, translation, and the ability to run Whisper entirely on-device without an internet connection. A Teams plan is available at $6 per seat per month, granting all Pro features to every invited member. The workflow is designed to minimize window switching. Users hold a configurable hotkey (default is the right Option key, but can be set to fn, Command, or Caps Lock), speak naturally, and release the key. The transcribed and cleaned text appears at the cursor position in any application. Speechcap also includes a transform menu that can be triggered by selecting text and pressing the hotkey, or by holding a modifier key while dictating and tapping a letter (for example, F for Formal, I for Improve, T for Translate). Additional features include custom vocabulary that learns user-specific spellings, local history with filtering by app, date, or transform type, voice-triggered text expansion for canned replies or signatures, and deterministic spoken-name substitution for emoji and punctuation (opt-in per name). The app also supports in-flight transforms, allowing users to apply rewriting or translation during dictation before the text is pasted. Speechcap is built with a focus on privacy and security. In Pro on-device mode, audio never leaves the Mac; transcription and cleanup are performed locally using Whisper and a small on-device language model. In the free tier and Pro cloud mode, audio is sent to a transcription provider over an encrypted connection and is discarded immediately after processing. The company states that it does not train on user data or share it with third parties. The application requires macOS and, for cloud modes, an internet connection. On-device mode works offline. Speechcap is intended for users who need fast, hands-free text input across any application, with the ability to refine and translate spoken content without leaving the current window.