Juno

Product Introduction

1. Definition: Juno is a local, open-source voice dictation and writing layer specifically engineered for macOS. It functions as an on-device speech-to-text (STT) engine with an integrated AI writing assistant, designed to capture spoken thought and convert it into clean, actionable text directly within existing applications.

2. Core Value Proposition: Juno exists to create the most natural and private voice-first writing workflow for Mac users. Its primary function is to eliminate the friction between thinking and typing, enabling users to dictate notes, emails, messages, and documents with live transcription, intelligent cleanup, and native system actions, all while ensuring absolute privacy through offline, on-device processing.

Main Features

1. Live Transcription HUD & Intelligent Cleanup: Juno provides a real-time, heads-up display (HUD) that shows words as they are spoken, offering immediate feedback. The core technical innovation is its AI-powered cleanup engine which processes raw speech input to automatically handle pauses, false starts, hesitations (e.g., "umm"), and corrections. It transforms fragmented speech into polished, grammatically sound text, formatting numbers, dates, currencies, and links (e.g., "nineteen ninety five" becomes 1995). This "cleaned by default" output reduces post-dictation editing.
2. Voice Commands & Mac-Native Actions: Users can invoke system-level actions using simple voice commands prefixed with "Hey Juno." This feature integrates directly with macOS services, allowing the creation of Apple Notes, Reminders with specific times, and Calendar Alarms from voice alone. For example, saying "Hey Juno, note this: prep the board meeting, remind me tomorrow at 9 to send the deck" files tasks to the correct native app.
3. Context-Aware Voice Rewrites: Juno allows users to select existing text within any application and use voice commands to rewrite or refine it in place. Through natural language instructions like "make this warmer," "shorter," or "more formal," the on-device AI rewrites the selected passage. This moves beyond simple dictation into active voice-driven editing and composition.

Problems Solved

1. Pain Point: The inefficiency and privacy risk of cloud-based dictation tools. Traditional voice-to-text services often require an internet connection, sending private audio data to external servers for processing, and may include subscription fees or usage caps. This creates latency, privacy concerns, and ongoing costs.
2. Target Audience: The primary users are Mac professionals (writers, developers, managers, researchers) who value productivity and privacy. This includes remote workers conducting sensitive communication, students taking fast notes, and creative professionals using voice to brainstorm. It's particularly valuable for users with repetitive strain injuries (RSI) or those who simply think faster than they type.
3. Use Cases: Essential scenarios include drafting confidential emails in Mail or Slack without data leaving the device, creating meeting notes in real-time within Notes or Obsidian, quickly logging reminders and alarms while on the move, and editing a document with verbal commands like "rewrite this paragraph to be more concise."

Unique Advantages

1. Differentiation: Juno fundamentally differs from Apple Dictation by adding a live HUD, intelligent AI cleanup, and voice commands. Compared to paid alternatives like Wispr Flow or Superwhisper, Juno's key differentiators are its complete open-source transparency, permanent free-tier with no usage cap, and exclusive focus on local, offline operation. It consolidates dictation, cleanup, editing, and system actions into a single local workflow, a combination not found in competing products.
2. Key Innovation: The pivotal innovation is the uncompromising local-first architecture. By running all transcription, AI cleanup, and command processing on the user's Apple Silicon Mac (M1 or later, macOS 14+), Juno guarantees that no audio, transcript, or usage data ever leaves the device after the initial one-time model download. This architecture delivers zero-latency response, ironclad privacy, and eliminates dependency on cloud services or subscriptions.

Frequently Asked Questions (FAQ)

1. How does Juno ensure my privacy for sensitive dictation? Juno is built on a fully local, on-device processing model. All audio capture, speech-to-text conversion, AI writing cleanup, and command execution happen directly on your Mac's Apple Silicon chip. No voice recordings, transcripts, or usage data are ever sent to external servers or used for training, making it inherently private for confidential work.
2. Is Juno just another dictation app, or does it do more? Juno is a comprehensive local voice layer that goes far beyond basic dictation. It combines real-time transcription HUD, AI-powered writing cleanup (removing fillers and formatting), voice-to-text editing of existing content, and Mac-native voice actions (creating Notes, Reminders, Alarms) into a single, integrated tool. Its goal is to be the primary interface for voice-driven work on Mac.
3. What are the system requirements, and is there an ongoing cost? Juno requires a Mac with Apple Silicon (M1 or later) running macOS 14 (Sonoma) or newer. 16GB of RAM is recommended for optimal performance. The product is free forever with no subscription tiers, usage caps, or account required. After a one-time local download of the speech models, it operates entirely offline.
4. How does the "Voice Rewrite" feature technically work? The rewrite feature leverages on-device large language models (LLMs) integrated into Juno. When you select text and give a voice command like "make this more formal," Juno processes the selected text and your instruction locally. The AI then generates a revised version of the text and uses system accessibility features to replace the original selection in your active application.
5. Does Juno work in all Mac apps like browsers, Slack, and documents? Yes, Juno is designed to work in any standard macOS text input field. This includes Mail, Notes, Messages, Slack, web browsers (Chrome, Safari), word processors, code editors, and AI prompt interfaces. For security, it automatically detects and blocks dictation into password fields or other secure inputs, showing a "Blocked" status to keep you in control.