Voxt: AI Voice Workflows for Mac
Voxt is a desktop voice input and transcription tool for macOS that turns speech into text and can use a language model to remove filler, fix punctuation, improve structure, and apply custom instructions, with local model processing, meeting transcription and analysis, translation, rewriting, and per-app enhancement rules.
About this tool
What
Voxt is Voxt is a desktop voice workflow app for macOS that compresses audio capture, speech recognition, cleanup, translation, insertion, history, and optional remote calls into the real writing surface of the Mac. Leave the cursor in the message, document, browser field, or editor, and start voice input directly.
Key capabilities
- Desktop voice input: Capture speech and convert it to text from any focused Mac input.
- AI transcription and enhancement: After transcription, a language model can remove filler, fix punctuation, improve structure, and apply your custom instructions.
- Per-app rules: Create different model enhancement rules for apps and websites, for example shaping responses as SQL while preserving the editor's expected syntax and format.
- Meeting transcription and analysis: Added editable meeting transcripts with waveform timelines, audio playback, and more durable meeting history; imported meeting file analysis supports bounded audio windows, full transcription, continuous speaker diarization, progress and cancellation, and history audio preference handling.
- Recording analysis: Analyze an existing recording on your Mac, with a current 12-hour meeting-file limit and queued background processing.
- Multiple capture modes: Capture system audio, microphone audio, or both with Meeting, Captions, or Recording mode.
- Global voice notes: Wake voice note capture globally instead of opening a notes app first.
- To-do status: A captured note can carry a to-do status so you can distinguish actionable items from general ideas in the floating note manager.
Processing
and privacy Voxt tells you what local processing covers, and it leaves room for the parts of the product that are not model inference: downloads, accounts, billing, analytics, support, and updates. Account features, billing, downloads, website analytics, and remote providers need a connection too. Actual processing time and resource use depend on the file, selected model, hardware, and whether speaker analysis is enabled.
Platforms
and availability Voxt supports macOS and Windows, uses a freemium model, and is open-source on GitHub.