VoxBee
VoxBee is an offline speech-to-text app for macOS and Linux that handles dictation, audio transcription, and meeting recording — all processed on-device. Dictate into any app with push-to-talk across 30 languages, with grammar correction and filler word removal. Transcribe audio files via drag-and-drop or URLs from 1,800+ sites including YouTube, Vimeo, and podcast RSS feeds. Record meetings from Zoom, Google Meet, Teams, FaceTime, and Webex with AI-generated summaries. Powered by 10 on-device models (WhisperKit and NVIDIA Parakeet), no internet, account, or subscription required — just a one-time purchase.
What it does
- •Push-to-talk dictation across all apps with 30 language support
- •Fully offline — 10 on-device Whisper and NVIDIA Parakeet models
- •Transcribe files or URLs from 1,800+ sites (YouTube, Vimeo, podcasts, and more)
- •Meeting recording with auto-detection of Zoom, Meet, Teams, FaceTime, and Webex
- •AI summaries via OpenAI, Anthropic, or local Ollama
- •Grammar correction, filler word removal, and personal dictionary with phonetic matching
- •Voice notes with AI transformation templates and screenshot capture
- •8 color themes, 8 waveform styles, and configurable hotkeys