Versions:
YazSes, published by MSKazemi and currently available at version 2.36.0 with two versions released overall, is a free, open-source offline voice dictation and speech-to-text tool designed for users who need transcription without any network dependency. Its core workflow is deliberately simple: hold a key, speak, and release, at which point the spoken words are transcribed locally on the user's own CPU, powered by faster-whisper, and typed directly into whichever window has focus. Because all processing happens on-device, the software belongs squarely in the offline speech recognition and privacy-focused productivity category, requiring no cloud service, no API key, no account, no subscription, and collecting no telemetry of any kind. This makes it well suited to situations where audio must not be sent to a third party, such as confidential meetings, air-gapped machines, or simply for users who prefer not to rent access to their own voice through hosted services. Beyond live dictation, YazSes can transcribe existing recordings through a command such as `yazses transcribe talk.m4a`, with optional speaker labels to distinguish participants, and it can capture an entire meeting into a speaker-labelled transcript accompanied by locally generated minutes, again without sending any data off the machine. Typical use cases therefore include private note-taking, drafting text hands-free in any application, documenting sensitive discussions, producing meeting records in restricted environments, and general speech-to-text work where confidentiality, cost, or connectivity constraints rule out cloud-based dictation platforms. As an open-source project, it offers an alternative to commercial dictation software for users who value transparency, local control of their data, and a one-time setup free of recurring fees, while still covering both real-time dictation and batch transcription of recorded audio in a single tool.
Tags: