ARMONI FOR MACOS SCREEN READER SOFTWARE

What Is Armoni?

Armoni is an open-source, AI-powered screen reader developed for macOS. It is being developed by WeAccess, a provider of AI-powered digital accessibility solutions, with the goal of offering a free alternative to VoiceOver, Apple's built-in screen reader for Mac.

The first beta version includes AI-powered image and video description features. These features use Apple's on-device Foundation Models (AFM) framework.

Why Was Armoni Developed?

The idea for Armoni emerged from the experiences shared by visually impaired employees at WeAccess regarding the transition from Windows to Mac. Employees who had used screen readers on Windows for years explained that changing their familiar navigation methods could make the decision to switch to Mac more difficult. They also noted that achieving the same level of efficiency in their daily tasks after switching could take time.

With VoiceOver, Mac's built-in screen reader, users navigate between elements by moving left and right. In the default navigation model, they must first interact with a group to access the elements within it. This navigation approach differs from the habits developed through using NVDA or JAWS on Windows.

In WebAIM's 2026 screen reader survey, 55% of respondents who answered the question about their primary desktop or laptop screen reader selected JAWS, 32.9% selected NVDA, and 6.5% selected VoiceOver.

WeAccess began developing Armoni to make this transition easier by bringing familiar Windows screen reader navigation patterns to macOS. The goal is to enable users, such as those familiar with NVDA, to start using a Mac with their preferred navigation methods after learning a few essential Mac-specific concepts.

The ability to run built-in language models directly on Macs with M-series processors also shaped Armoni's design. Apple's Foundation Models framework made it possible to integrate AI capabilities into the screen reader without relying on an external service.

In its first beta release, Armoni provides descriptions of photos and videos. It also generates meaningful labels for buttons and other controls that do not have accessible names.

In future development stages, WeAccess aims to introduce AI agents capable of performing tasks within applications based on user instructions. Another goal is to develop an AI-powered AX repair tool that will make it easier to create scripts for addressing accessibility issues in applications.

What Kind of Experience Does Armoni Aim to Provide?

Armoni aims to offer a familiar keyboard experience inspired by Windows screen readers while preserving macOS's native interface and functionality. One example is using Tab to reach a list, navigating through its options with the arrow keys, and making a selection. The same navigation approach extends to essential Mac areas, including the desktop, Dock, menu bar, context menus, and status menus.

In Finder, users can interact with Mac-specific areas such as the sidebar and toolbar. They can browse files and folders in different view modes using the arrow keys, open a selected item with Enter, and move to the parent folder with Backspace.

Searching in Spotlight and navigating through results provides a familiar workflow for users accustomed to Windows Start menu search. Armoni combines this familiar experience with Spotlight's Mac-specific search capabilities.

On the web, Armoni offers a navigation experience similar to what JAWS users are familiar with. It reads information such as a button's name or whether a checkbox is checked from the accessibility tree (AX) exposed by the browser. When interpreting this information, Armoni follows WAI-ARIA role and state definitions. WCAG success criteria and ACT test rules also provide a common reference framework for evaluating the web experience.

What Can Armoni Do Right Now?

Mac Navigation and File Operations. You can access the desktop, Dock, menus, and Notification Center using the keyboard, and switch between the main regions of a window with F6. In Finder, you can browse files in different views, select multiple files, and copy or move them. You can also read Spotlight search results and open the selected result.

Split Application Views. Armoni allows you to move between lists, details, and composition areas within the same window. In the Phone app, you can navigate between recent calls and contact details. In Messages and WhatsApp, you can move between the conversation list, message history, and message composition area. In Mail, you can access mailboxes, message lists, and the body of an open message separately.

Codex and Claude. In the Codex Mac application, you can switch between the sidebar, conversation transcript, and composition area. In the Claude Mac application, you can navigate through the content area containing the conversation.

Virtual Document Navigation in Content Storefronts. In supported content areas of the App Store and Apple TV, you can use a virtual document cursor to read storefronts as if they were web pages. You can navigate through card names and details using the arrow keys and activate a selected item with Enter. In Apple TV, a separate interaction command provides access to the storefront's page navigation controls.

Reading and Interacting With Web Pages. In Safari and Chrome, you can read pages using the arrow keys in browse mode and jump to headings, links, buttons, and tables with single-key commands. In focus mode, you can interact with form fields. When navigating between table cells, you can hear row and column headers and copy selected text from the page.

Reading Word Documents. You can read document text, navigate through headings, lists, and tables, and hear formatting information such as bold or italic text while reading. Separate commands allow you to check your position in the document, the formatting of selected text, and footnote information. Some commands require macOS automation permission.

Text Search, Continuous Reading, and Translation. With Armoni Find, you can search for a word on web pages and in supported application text. You can pause continuous reading and resume from where you left off. If the required language packs are installed, you can translate selected English text into Turkish and listen to the translation.

Text Recognition and Image Descriptions. Using OCR, you can recognize text in an image of the screen, active window, or focused element and have the results read aloud. You can request descriptions of photos and screenshots. For supported video sources, you can request a description of the current frame or receive scene descriptions during playback.

Labeling Unnamed Controls. With the Labeler, you can assign your own label to a supported button or other control, or ask Armoni to generate one. AI-powered label generation and visual descriptions run on-device using a compatible Apple model and require screen recording permission.

Speech Preferences and Settings. In the first beta, you can adjust the voice, language, speech rate, and level of punctuation detail. You can also use the speech dictionary to change how specific words are pronounced. Armoni's data-driven architecture provides the foundation for managing general preferences and application-specific settings from a single interface in future releases.

What Should You Expect From the First Beta?

The priority of the first beta is to explore how everyday tasks work with Armoni across macOS's main interface areas and built-in applications. You can begin by trying a few tasks you frequently perform on your Mac, then move on to other applications you use.

1. Core Mac Areas and Finder. Try navigating the desktop, Dock, menu bar, status menus, and context menus. Perform a Spotlight search and open a result. Access frequently used controls in Control Center, Notification Center, and System Settings. In Finder, open a folder, navigate between files, and select an item.
2. Built-in Mac Applications. Browse content in an App Store or Apple TV storefro