Star Growth
Overview
Jarvis AI Assistant is an open-source voice dictation tool for Mac that converts speech to text using a simple key-hold interface. Users hold the Fn key, speak, and release to generate clean, punctuated text anywhere on their system. The tool automatically removes filler words like 'um' and 'like', fixes grammar, and can rephrase or format text according to customizable prompts. It supports both offline operation using local Whisper or NVIDIA Parakeet models via Sherpa-ONNX, and cloud-based processing with Deepgram and Gemini APIs. Beyond dictation, Jarvis can perform basic voice commands like opening applications or setting timers. The application emphasizes privacy with no telemetry or tracking, and offers full prompt engineering capabilities to customize behavior for different use cases like email formatting or general text cleanup. Available for both Apple Silicon and Intel Macs, with iOS support in TestFlight beta, Jarvis is signed and notarized by Apple for seamless installation.
Deep Analysis
vs Wispr Flow ($10-24/month): 100% free, open-source, fully offline-capable voice dictation with zero telemetry
⚡ Capabilities
- • Voice-to-text dictation with one-key activation
- • Automatic filler word removal and grammar correction
- • Fully offline mode with local Whisper or NVIDIA Parakeet
- • Local LLM support via Ollama
- • AI post-processing (rephrase, bullet-point, generate)
- • Custom voice commands and prompt engineering
🔗 Integrations
✓ Best For
- ✓ Mac users wanting free, private voice dictation
- ✓ Developers who want offline-capable speech-to-text without subscriptions
✗ Not Ideal For
- ✗ Windows or Linux users
- ✗ Enterprise-scale voice transcription pipelines
Languages
Deployment
⚠ Known Limitations
- ⚠ macOS only (Windows version not yet available)
- ⚠ Local LLM adds 1-3s latency on standard M1/M2 chips
- ⚠ iOS version still in TestFlight beta
- ⚠ Voice commands are basic (open apps, set timers)
Pros
- + Completely free and open-source with no subscription fees or hidden costs
- + Works fully offline with local AI models for privacy and independence from cloud services
- + Highly customizable through prompt engineering to adapt behavior for different text formatting needs
Cons
- - Limited to Mac and iOS platforms with no Windows or Linux support
- - Requires manual setup and configuration of AI models for optimal performance
- - Voice command actions are basic compared to full virtual assistant platforms
Use Cases
- • Writers and content creators who need fast, accurate voice-to-text conversion without filler words
- • Privacy-conscious users requiring offline dictation for sensitive documents or communications
- • Professionals who frequently switch between typing and speaking for email composition and note-taking