VoxBee

VoxBee provides on‑device speech‑to‑text capabilities for macOS and Linux, allowing users to dictate, transcribe audio files, and capture meeting conversations without sending data to external services. It runs Whisper and NVIDIA Parakeet models locally, supporting thirty languages and offering grammar correction, speaker diarization, and AI‑generated summaries that can be exported in markdown.

The application works across any foreground program via a push‑to‑talk hotkey, letting users speak while writing emails, code, or notes. Audio files or URLs from over a thousand sources can be dropped into the interface for automatic transcription, and meetings on platforms such as Zoom, Google Meet, or Teams can be recorded by capturing both system audio and microphone input.

A single dashboard aggregates dictation statistics, waveform visualizations, personal dictionaries, and configurable themes, keeping all voice‑related work in one place while ensuring that every word remains on the user’s device.