Product Introduction
Definition: Megaphone is a free, open-source, native macOS dictation application that leverages Apple’s on-device SpeechAnalyzer for real-time transcription and on-device Foundation Models for smart cleanup, app-aware context adaptation, and inline voice commands. It is a productivity tool classified as a speech-to-text and voice dictation utility, designed exclusively for Apple silicon Macs running macOS 26 or later. The application is MIT-licensed, requires no account, API key, subscription, or server, and processes all audio locally on the user’s machine.
Core Value Proposition: Megaphone exists to provide a private, free, and highly intelligent dictation experience that eliminates the need for cloud-based services. It combines fast, local transcription with cutting-edge on-device AI to remove filler words, correct self-corrections, adapt vocabulary and tone to the active application (e.g., Mail, Slack, Terminal), and execute inline AI commands—all without sending voice data to any external server. This addresses the growing demand for privacy-preserving, offline-capable voice input that matches the productivity of premium subscription tools.
Main Features
Smart Cleanup with On-Device Foundation Models: Megaphone uses Apple’s on-device language models to automatically remove filler words (e.g., “um,” “uh”), repetitive stutters, and self-corrections (“Thursday—no, Wednesday”). The cleanup operates in real time as the user dictates, with a deterministic fallback for speed. The system also preserves technical syntax (e.g., URLs, code identifiers) when appropriate, demonstrated by bounded corrections that avoid altering non-speech patterns. Users can choose between Smart, Basic, or Exact cleanup modes, with Smart mode providing the highest AI-powered fidelity.
App Context & Screen Awareness: The application reads the active app and window title locally to adapt tone, spelling, and formatting. For example, dictation into a Mail draft yields warmer, conversational language, while the same voice input in Slack is automatically rendered concise. In Terminal, Megaphone preserves command-line flags and case-sensitive terms. This context adaptation is powered by Apple Foundation Models running entirely on the Neural Engine, ensuring no latency or privacy leak.
Inline AI Commands with “Hey Megaphone”: Users can begin an instruction with “Hey Megaphone” (e.g., “Hey Megaphone, make that formal” or “Hey Megaphone, format that as bullets”) and the on-device Foundation Model rewrites the dictated text directly at the cursor—no separate chat interface or copy-paste required. The command is processed using the current app context and the user’s own voice input, enabling fluid, hands-free editing and formatting without any cloud dependency or API key.
Personal Dictionary & Automatic Learning: Megaphone includes a native Dictionary settings panel where users can add names, technical terms, acronyms, and product names to ensure correct spelling during transcription and cleanup. The system also features automatic learning: words used successfully across three dictation sessions are added to the user’s vocabulary. The Dictionary supports import/export via JSON files, allowing users to synchronize their private vocabulary across multiple Macs. This ensures consistency for specialized jargon (e.g., “Kuber”, “SpeechAnalyzer”, “Megaphone”).
Multi-Language Support: The app detects and switches between languages naturally—English, Español, Français, हिन्दी, Português, and more—without manual toggling. The on-device SpeechAnalyzer handles multiple simultaneous languages, making it suitable for bilingual or multilingual professionals.
Problems Solved
Pain Point: Traditional cloud dictation services (e.g., Otter.ai, Dragon) require API keys, subscriptions, and internet connectivity, raising privacy concerns and recurring costs. Megaphone solves this by offering a completely offline, zero-cost solution that keeps all voice data on the device, eliminating the risk of cloud leaks and the need for ongoing payments.
Target Audience: Software developers, system administrators, technical writers, content creators, academics, and professionals who work in Mac-native environments (Mail, Slack, Obsidian, Terminal, IDEs like Cursor, Claude Code) and require high-accuracy dictation that respects technical syntax and app-specific tone. It is also ideal for privacy-conscious users who cannot use cloud-based dictation due to corporate or regulatory compliance (e.g., healthcare, legal, defense).
Use Cases: Writing release notes while keeping hands on the keyboard; dictating code comments or git commands in Terminal without breaking flow; composing professional emails in Mail with a warmer tone; drafting bullet points in Obsidian or Slack with concise formatting; performing inline rewrites and formatting commands without switching windows or using a mouse; teaching the system custom industry terminology (e.g., product names, SDK names) for error-free transcription.
Unique Advantages
Differentiation: Unlike competing dictation tools (e.g., macOS built-in dictation, Dragon Dictate, Otter.ai), Megaphone combines four unique elements: (a) on-device smart cleanup using Apple Foundation Models (not simple rule-based removal); (b) app-aware context adaptation that changes tone and formatting per window; (c) inline AI commands that operate without a separate chat interface; and (d) a completely free, open-source (MIT) license with no subscriptions, accounts, or cloud dependencies. Competitors typically require monthly fees, send data to the cloud, or lack contextual awareness.
Key Innovation: The use of Apple’s on-device SpeechAnalyzer for live transcription paired with Foundation Models (running on the Mac’s Neural Engine) for intelligent editing is a novel architectural choice. This allows both real-time accuracy and sophisticated post-processing without latency or network calls. Additionally, the “Screen Awareness” feature that reads the active window title and app bundle ID to infer tone and formatting is a pragmatic implementation of local context inference, not available in any other dictation tool.
Frequently Asked Questions (FAQ)
Does Megaphone require an internet connection or API key to work?
No. Megaphone runs entirely on your Mac using Apple’s on-device SpeechAnalyzer and Foundation Models. No account, API key, subscription, or internet connection is required. All voice processing happens locally on the Neural Engine, ensuring complete privacy and offline functionality.Which macOS versions and hardware are supported?
Megaphone is compatible only with macOS 26 or later and requires Apple silicon (M-series chips). It does not support Intel-based Macs. The app is distributed as a native DMG installer and is available for free under the MIT open-source license.How does Megaphone handle custom vocabulary and technical terms?
Users can add words to the Dictionary via the settings panel. The system also learns frequently used terms automatically after three successful dictations. The Dictionary can be exported and imported as a JSON file, allowing you to migrate your vocabulary to another Mac. This is particularly useful for names, product names, acronyms, and programming syntax.Can I use Megaphone to give commands like “make this formal” or “list as bullets”?
Yes. By beginning a dictation with “Hey Megaphone” followed by an instruction, the on-device Foundation Model will interpret your request and rewrite the text directly at the cursor. This works in any app, and the model considers the current app context (e.g., it will produce bullet points in a note-taking app or formal language in an email). No additional window or copy-paste is needed.Is Megaphone truly free and open-source? What is the license?
Yes. The app is entirely free, with no hidden costs, in-app purchases, or subscriptions. It is released under the MIT open-source license, and the source code is available on GitHub. The DMG can be downloaded from the official website or installed via a single Terminal command.
