VoiceBean is launching soon — we're getting payment integration ready.
100% Local & Private

Privacy-first voice input,
powered by local AI.

Fast, precise voice input for macOS. Multiple local recognition engines with on-device LLM correction — everything stays on your Mac.

See Features
Your voice stays on your device. Always. →
AI Smart Correction
100% Local

Core Features

Built for professionals

Everything runs locally. No cloud, no latency, no compromise.

Real-time Transcription

Whisper-powered recognition with interim results streamed as you speak. Supports Chinese, English, Japanese, Korean and more.

AI Error Correction

Local LLM (Qwen / Llama / Gemma) automatically fixes homophones, punctuation and phrasing after transcription.

Privacy First

All processing happens on your Mac. Audio and text never leave the device — no network required.

Smart Scene Detection

Automatically adapts correction style based on the active app — coding, writing, email, chat or search.

Global Hotkey

Hold Fn, Command, Option or Control to record. Release to commit. Works system-wide in any text field.

Multiple Engines

Choose from Apple Speech, Whisper, SenseVoice or Paraformer. Pair with llama.cpp or Apple MLX correction.

Power Features

Beyond voice input

Powerful features that make VoiceBean more than a dictation tool.

Smarter voice input

Not just transcription — translate, filter, and refine your voice input.

Transcribe

Speech to text with optional AI correction.

Translate

Shift + hotkey — transcribe and translate to your target language.

Filler word removal

Automatically removes "um", "uh", "like", "you know" for cleaner output.

5 recording indicators

Floating ring, screen-edge glow, Dynamic Island pill, spatial bubble, or three-body animation — pick the style that fits your workflow.

Custom dictionary

Add domain-specific terms, brand names, and jargon. Injected into both recognition and AI correction for higher accuracy.

Encrypted iCloud sync

Sync usage stats and transcription history across devices with end-to-end encryption via CryptoKit. Your data stays private even in the cloud.

Lab: custom model import

Import your own GGML (Whisper) or GGUF (correction) models. Full control for power users and researchers.

Your voice stays on your device. Always.

VoiceBean is designed from the ground up with privacy as a core principle. No audio leaves your Mac, no text is sent to any server, and no account is needed to use core features.

Zero audio upload

Your voice recordings are processed entirely on-device and never sent anywhere.

Works fully offline

No internet connection needed. Recognition and correction run locally on Apple Silicon.

Local-only data

History, settings and models are stored in your Mac's app sandbox. We have no access.

Design Philosophy

Focus on speaking, not staring at the screen

Most voice input tools display text as you speak. It seems efficient, but actually splits your attention — you can't help reading the jumping words on screen, losing your train of thought. VoiceBean takes a different approach: just focus on what you want to say. AI handles recognition, correction, and punctuation in the background, then delivers the complete result all at once. Your thinking stays uninterrupted, and the text quality is higher.

The problem with streaming
The words I am saying right now would appear one by one and you would keep glancing at them|

Text appears word by word → attention diverted → train of thought broken → expression quality drops

VoiceBean
Everything I'm saying right now will be recognized and corrected by AI in the background, then output as a complete, punctuated result all at once.

Focus on speaking → AI processes in background → complete result at once → uninterrupted flow

Pricing

Simple pricing

Start free, upgrade when you're ready.

Free

Try basic voice input

$0
  • Apple Speech recognition
  • Basic voice input
  • Basic model downloads
  • 50 history records

Pro Monthly

Flexible month-to-month

$5.9 / mo
  • Everything in Free
  • Unlimited Whisper / SenseVoice / Paraformer
  • AI correction (llama.cpp & MLX)
  • All model downloads
  • Smart scene detection

Pro Annual

Best Value

Best value plan

$29.9 / year
Save 58% vs monthly
Just $2.5/mo
  • Everything in Free
  • Unlimited Whisper / SenseVoice / Paraformer
  • AI correction (llama.cpp & MLX)
  • All model downloads
  • Smart scene detection
  • Priority support
Lifetime

Pro Lifetime

Pay once, use forever

$199 one-time
  • Everything in Free
  • Unlimited Whisper / SenseVoice / Paraformer
  • AI correction (llama.cpp & MLX)
  • Permanent access, no renewals
  • Free updates for future versions
  • Priority support

Download

Download VoiceBean

Free download, instant local AI voice input. Everything works out of the box, no internet required.

Requires Apple Silicon (M-series chip) Mac

System Requirements

Apple Silicon

M1 / M2 / M3 / M4 / M5

macOS 15.0+

Sequoia or later

~200 MB

App only, models downloaded separately

Ready to transform your voice input?

Download VoiceBean and start dictating with local AI today.