Versions:
Whisperstream is a dictation application for Windows developed by Lanreal Technologies Inc., currently available in version 1.0.0, which is the sole release listed in the catalog. The software belongs to the speech-to-text and voice input category, with a particular emphasis on privacy and on-device processing. Its core purpose is to convert spoken words into written text quickly and privately, without sending audio to external servers. The application operates through a push-to-talk model: the user holds a designated key, speaks, and the resulting text is typed directly into whatever window currently has focus. This design makes it suitable for a wide range of everyday use cases, including composing emails, writing documents, filling out forms, chatting, and entering text into any application that accepts keyboard input, all without switching tools or copy-pasting from a separate transcription window. A defining characteristic of Whisperstream is that all transcription runs locally on the user's CPU using NVIDIA Parakeet, meaning audio never leaves the machine. This local-first architecture distinguishes it from cloud-based dictation services and makes it appropriate for users who handle sensitive information or simply prefer that their voice data remain under their own control. Beyond basic transcription, the software includes a custom dictionary, which allows users to teach the system specialized vocabulary, names, or terminology relevant to their work. It also offers optional AI cleanup, which can refine raw transcriptions into more polished text, and a searchable transcript history, enabling users to revisit and retrieve past dictations when needed. Together, these features position Whisperstream as a productivity-oriented dictation tool for Windows users who want fast, accurate speech-to-text input that works anywhere on the system while keeping the entire processing pipeline private, offline, and fully contained on the local device.
Tags: