OpenASR

What OpenASR is

OpenASR converts speech into text on the user’s computer for recordings, dictation, and live captions. Audio, models, and transcripts stay on the device, and transcription can continue without a network connection. The product says it has no telemetry, analytics SDK, account, or cloud transcription route; optional text refinement or translation sends recognized text to the chosen provider. The desktop app handles file transcription, dictation into other applications, and live captions, with speaker separation and timestamps. Exports include plain text, structured data, and timestamped subtitles. Users can choose a model from a catalog, and models run locally. Desktop apps are available for Apple Silicon Macs and Windows 10 or later on x64; a native Linux desktop app is planned. The public Apache-2.0 core includes a CLI, local HTTP API, model format, and Docker server, with Linux support for CLI and server use. The local server supports part of the OpenAI audio transcriptions API. The first-use model download is roughly 1 GB, and installed model packs stay on disk until removed. OpenASR Desktop is listed as free.