AriaType
Privacy-first voice keyboard with AI polish (Beta, macOS)
Last verified:
What is AriaType?
AriaType is an open-source, local-first voice-to-text input application for macOS and Windows that serves as a voice layer for your desktop. It turns spoken thoughts into context-aware text that appears directly where your cursor is positioned, allowing you to reply, take notes, draft prompts, and clean up documents without leaving the app you're already using. The tool stays in the background and uses a global hotkey (default Shift+Space) to activate voice input—hold the key, speak naturally, release, and your words appear in the active application.
Key features include dual local STT engines (Whisper and SenseVoice) that you can switch between based on language and accuracy needs, support for 100+ languages with automatic detection, and intelligent text polishing that removes filler words, adds punctuation, corrects grammar, and adjusts tone. The app offers template-based polishing with presets like Remove Fillers, Formal Style, Make Concise, and Agent Prompt, plus optional cloud services for enhanced accuracy. It includes a customizable capsule floating window showing recording status, local history with search, usage statistics dashboard, audio preprocessing with noise reduction and VAD静音裁剪, and stable text injection via keyboard simulation with clipboard restoration.
AriaType is designed for software developers who want to dictate documentation, commit messages, and code comments without leaving their IDE; content creators and writers capturing ideas at the speed of thought; privacy-conscious professionals like lawyers and medical researchers handling confidential information; accessibility users with RSI or motor impairments; and anyone who wants to speak faster than they type in applications like VS Code, Slack, Notion, browsers, Mail, and Obsidian.
The tool prioritizes privacy with local-first processing by default—speech data stays on your device and never uploads to the cloud unless you explicitly enable optional cloud services. It supports domain-specific enhancement with glossaries and custom terminology for IT, legal, and medical scenarios, and provides model recommendations based on your language choices.
AriaType pricing
Pricing model: Freemium
AriaType is free and open-source under AGPL-3.0 license. No subscription or account required for core features. The v0.1 version is completely free to download and use immediately. Optional cloud services (Cloud STT and Cloud Polish) can be enabled but require separate configuration with your own cloud service credentials. There is also mention of a free tier with 5 daily credits and paid tiers at $20/$50/$100 for certain cloud-enhanced features, though the core open-source application remains free.
AriaType pros
- Open-source with AGPL-3.0 license for full transparency
- Local-first architecture ensures privacy by default
- Global hotkey workflow (Shift+Space) for seamless voice input
- Direct-to-cursor text insertion without copy-paste
- Dual local STT engines: Whisper and SenseVoice
- Supports 100+ languages with automatic detection
- Chinese and CJK optimization with SenseVoice model
- Intelligent text polishing removes fillers and adds punctuation
- Four built-in polishing templates plus custom templates
- Optional cloud STT and Cloud Polish for enhanced accuracy
- Domain-specific glossaries for professional scenarios
- Local history with search for reusing common expressions
- Usage statistics dashboard tracks productivity metrics
- Customizable capsule floating window with real-time status
- Audio preprocessing with noise reduction and VAD silence trimming
- Works across all macOS and Windows applications
- Supports macOS 12+ and Apple Silicon for low latency
- No account or subscription required for core features
- Streamed intermediate results from cloud STT providers
- Model download and status management built-in
AriaType cons
- Currently optimized primarily for macOS (Windows support newer)
- Requires macOS 12 or later, not compatible with older Macs
- Local model downloads required for offline functionality
- Apple Silicon recommended for best performance
- Open-source means no official customer support
- Cloud services must be manually configured and enabled
- Steeper learning curve for custom template setup
- May have higher system resource usage during local processing
Frequently asked questions about AriaType
Is AriaType completely private and offline?
Yes, AriaType is designed with a local-first philosophy. By default, it uses local speech recognition models that process your voice data on your Mac's CPU/GPU. Audio data remains on your local SSD and never leaves your device. Cloud services are strictly optional and must be explicitly activated by the user in the settings.
How does AriaType insert text into other apps without copy-pasting?
AriaType uses macOS accessibility APIs to simulate keyboard input at the current cursor position. Once audio processing completes, the application types the text directly into whichever app is currently in focus, whether it's a web browser, terminal, word processor, Slack, VS Code, or any other application.
Does AriaType require a subscription or account?
No, AriaType v0.1 does not require an account or subscription. It is an open-source tool that can be downloaded and used immediately. This aligns with the product's philosophy of focusing on utility over user acquisition metrics.
Which versions of macOS are compatible with AriaType?
AriaType is optimized for modern macOS environments and requires macOS 12 (Monterey) or later. It leverages recent advancements in Apple Silicon and macOS frameworks to ensure low-latency transcription and smooth background operation.
What is the difference between Whisper and SenseVoice models?
Whisper offers broader language coverage with multiple model tiers, making it ideal for English and cross-language scenarios. SenseVoice is optimized for Chinese, Traditional Chinese, Cantonese, and CJK use cases, providing better accuracy for Mandarin speakers. The app recommends models based on your selected language.
How do I customize the hotkey for voice input?
AriaType supports custom hotkey configuration in the desktop settings. The default hotkey is Shift+Space (hold to record, release to insert). You can modify this to any combination including Fn key support, and choose between hold-to-record or toggle recording modes.
Can I use AriaType with cloud services for better accuracy?
Yes, AriaType offers optional Cloud STT and Cloud Polish services that can be enabled separately in the Cloud Services settings. This gives you flexibility to use local processing by default while accessing cloud-enhanced accuracy for specific tasks. Streamed intermediate results are supported from cloud STT providers.
What polishing templates are available?
AriaType includes four built-in polishing templates: Remove Fillers (removes um/ah/well), Formal Style (makes text more professional), Make Concise (compresses expressions), and Agent Prompt (optimizes for AI prompts). You can also create custom templates for your specific needs.
How do I set up domain-specific terminology?
AriaType supports domain and subdomain settings plus a glossary feature for professional scenarios. You can configure initial prompts and terminology lists for IT, legal, medical, or any specialized field. This helps the model recognize industry-specific terms and improve accuracy in your domain.
Where are my transcription histories stored?
All transcription results are saved locally on your device and can be searched and reviewed through the local history feature. This makes it easy to reuse common expressions and maintains privacy since history never syncs to the cloud unless you explicitly configure external storage.