Megaphone

Megaphone is an open-source dictation application designed for macOS that operates entirely on-device, ensuring user privacy and offline functionality. It allows users to hold the Fn key, speak naturally, and have their words transcribed into clean text directly within any active Mac application. The core of Megaphone's functionality lies in its integration with Apple's advanced on-device technologies, specifically SpeechAnalyzer for real-time transcription and Foundation Models for intelligent text processing. The problem Megaphone addresses is the common reliance on cloud-based dictation services, which often come with privacy concerns, subscription fees, and the need for an internet connection. Many existing solutions involve sending audio data to external servers for transcription and processing, creating potential bottlenecks and security risks. Megaphone offers a compelling alternative by leveraging the power of modern Apple hardware and software to provide a seamless, private, and efficient dictation experience. One of the key features of Megaphone is its use of Apple's SpeechAnalyzer API, which transcribes spoken words in real-time as the user speaks. This ensures immediate feedback and a fluid dictation process. Complementing this is the integration of on-device Foundation Models. These models are responsible for a sophisticated "Smart Cleanup" process that automatically removes filler words (like "um" and "uh"), corrects self-corrections made by the user, and intelligently adds punctuation, resulting in cleaner and more readable text output. Megaphone also offers "Inline AI" capabilities, allowing users to initiate text generation or rewriting directly at their cursor's location by using a "Hey Megaphone" command. This feature extends beyond simple transcription to provide AI-powered text assistance within any application. Furthermore, the application is designed to adapt its formatting and vocabulary to the specific context of the active application. For instance, when used in an email client, it will adopt an email-like writing style, and when used in a messaging app, it will adjust to that platform's conventions. To enhance accuracy and personalization, Megaphone includes a private "Dictionary" feature. Users can teach the app to recognize specific names, acronyms, and technical terms, ensuring that these are transcribed correctly. This dictionary is stored locally on the user's machine, maintaining privacy. The application also supports multiple languages for transcription and offers various input modes, including hold-to-talk, toggle recording, and custom shortcuts, along with the ability to create voice macros for repetitive tasks. The overall workflow of Megaphone is built around privacy and efficiency. By processing all audio and text data locally on the user's Mac, it eliminates the need for external servers, API keys, or subscriptions. The application is developed natively in Swift for Apple silicon Macs running macOS 26 Tahoe, ensuring optimal performance and integration with the Apple ecosystem. It builds upon the FreeFlow project, replacing its cloud transcription components with Apple's on-device solutions. The primary benefits for users include enhanced privacy due to 100% on-device processing, improved efficiency through faster and cleaner transcriptions, and cost savings by being free and open-source. The adaptive formatting and personalized dictionary further contribute to a more tailored and accurate dictation experience. Concrete use cases for Megaphone include taking meeting notes directly into a document, dictating emails and messages with context-aware formatting, drafting code snippets in a terminal or IDE, writing reports, and even composing creative content, all without leaving the application where the text is needed. Megaphone is targeted at Mac users who value privacy and efficiency in their workflow. It is an open-source project licensed under MIT and is native to Apple silicon Macs running macOS 26. The project is actively maintained, with its source code publicly available on GitHub. In summary, Megaphone provides a powerful, private, and free dictation solution for Mac users by leveraging advanced on-device Apple technologies for accurate transcription and intelligent text processing.