Source-verified reference
Setup, every one-to-four-finger gesture, rotor controls, context menus and all settings groups, with their actual names and locations.
11 September 2026 · two release channels
Explore formatting analysis in the stable production channel and Speech history with a text-only AI flow in Beta.
Production
Receive the font, size, style, underline and colour data exposed for focused text through speech or sound. Optional screen sampling assists colour analysis, while accessibility IPC and focus paths are hardened against freezes.
Beta
Open the last 10 utterances with time and category information, copy an item, ask AI using text-only context, and hear a dynamic estimated percentage while the response is prepared.
Recorded evidence for both signed packages covers release identity, 29 locale resources, Direct Boot reboot, screen-off/Doze/keyguard cycles, service binding and touch exploration. AI results still depend on the selected provider and data exposed by the application.
114-9 Production · 114-10 Beta
All 15 main settings groups are covered in 14 chapters by combining Training and help with About in the final chapter. The 114-9 and 114-10 Beta features appear inside their owning settings sections.
Includes narrated explanatory slides, selectable captions and a full transcript; it is not a recording of every operation on a physical device.
Download the English transcript
Speech volume controls the screen reader volume. Text-to-speech settings select the main engine, language, voice and speed. Enable a separate notification synthesizer to choose its engine and voice independently. Battery and charging announcements have their own voice, percentage and time intervals, low battery, connection and full-charge information. Verbosity determines which details are spoken. Sound and vibration configures feedback and the sound profile. An engine must be installed on the phone; appearing in the engine list does not guarantee successful speech. Open Audio and find the main engine selection. Check that notification and battery announcements use separate engine choices. You can listen to the battery preview. Inspect percentage and time intervals; restore any unwanted changes before returning to training.
Open this chapter in the videoKeyboard layout, number row, letter size, alternate-character delay, volume-key cursor control and capitalisation personalise typing. Keyboard appearance has its own theme, height, width and custom colours, separate from Interface settings. Sound and vibration feedback levels can be adjusted. Word suggestions, spoken suggestions and a personal dictionary support typing. To use the keyboard, enable and select it in Android input methods. Open keyboard settings and find the number row and letter size. Read whether the number row is enabled. Optionally change it and restore its previous state. Check that the keyboard theme is separate from the settings screen theme.
Open this chapter in the videoThe notification filter can allow all apps or use allowed and blocked app lists. Entries are interpreted according to the selected filter mode. Time announcements offer an interval, speech or sound effect, effect selection, engine, voice and preview. Quiet hours use start and end times to restrict announcements. Reading notifications requires Android notification access. The separate notification voice is in Audio; live notification testing is in Advanced settings under Automatic diagnosis. Find the filter mode and quiet hours. Without changing the filter, consider which apps can be read. If time announcements are enabled, try their preview. Check how quiet hours can cross midnight, then return with Back.
Open this chapter in the videoCall management includes one-finger call controls, volume control during calls and automatic speakerphone. These options depend on the current call state. People nearby may hear a call on speakerphone. Call controls require appropriate phone and screen-reader state. Read each description and choose options that suit your daily use. Open call settings and read the current state of all three switches. Return when you can explain the effect of automatic speakerphone. This exercise does not require starting or ending a real call.
Open this chapter in the videoFocus indicator settings choose its thickness and colour. Gestures manages existing assignments, custom gesture creation and saved layouts. Volume-key shortcuts have separate settings. Customise menus manages AccessiMind, reading and navigation menu items. Proximity sensor behaviour and automatic icon or image descriptions are also here. Before changing a gesture, read its current action; avoid accidentally replacing a gesture you rely on to return. Open Customise menus, then inspect AccessiMind menu options. Find whether an item is shown. Optionally change it and restore its previous state. Inspect gesture assignments to identify your current double-tap and navigation actions without reassigning them. Release 114-10 Beta adds Speech history to the three-finger AccessiMind Reader menu. Review the last ten spoken utterances with time and category information, copy an item, or ask AI about it. Its visibility is controlled independently under Customize menus.
Open this chapter in the videoThe translation switch enables the feature. Engine, source language, target language and copying to the clipboard are separate choices. Required offline language models can be downloaded. With an online engine, text may be sent to its service. Correct source and target languages affect the result. Downloading a model does not make a different online engine work offline. Open the source and target language lists and read their current selections. Cancel without choosing a new value. Explain the clipboard option. This exercise requires neither sending text nor downloading models.
Open this chapter in the videoAI settings manage the selected provider, model and connection configuration. Screen or image descriptions depend on device support, provider support and permissions. Agent mode can divide a task into steps; inspect the actions and confirmations shown. Cloud features send content to the service. Do not share private information such as API keys. Selecting a model alone does not prove a working connection or a completed task. Open AI settings and locate the provider and model. Read their descriptions. Inspecting the configuration does not require sending a screenshot or running an Agent task. Return without changing private fields. Ask AI in Speech history sends only the selected spoken text as dedicated context, rather than a screenshot. Choose Explain or type a question. Release 114-10 Beta announces an estimated percentage that rises while the selected model prepares a response; it is waiting information, not a guarantee that the task will finish.
Open this chapter in the videoCustom voice commands connect a spoken phrase to a chosen action. You can inspect, add and edit saved commands. Keep phrases short and distinct. Recognition requires microphone permission and a suitable recognition service. A saved command does not guarantee recognition in every environment; test a safe action in a quiet place. Open the command list and inspect an existing phrase and its action. If the list is empty, open the add screen, read its fields and leave without saving. Do not use deletion or message-sending actions for practice.
Open this chapter in the videoPer-app profiles customise supported screen-reader preferences for a particular app. Select the app first and check the title to confirm which profile you are editing. Profile preferences are distinct from general preferences. Do not assume every setting supports profiles. Evaluate the effect when switching between apps. Open profiles and locate app selection. If inspecting a profile, check the app in the title. Return without saving or deleting. Mark the step when you can explain the difference between general settings and an app profile.
Open this chapter in the videoTools contains the AccessiMind browser and app shortcuts. The browser supports web navigation. App shortcuts manage saved operations; inspect the destination and steps before running one. Automation depends on the target app state and permissions. Do not use banking, purchases or message-sending flows for training. Open Tools and find the browser and app shortcuts. Inspect the shortcut list, then return without running an automation. Check that viewing a shortcut and running it are separate operations.
Open this chapter in the videoThe Braille keyboard and Braille display are separate ways to interact. The keyboard supports Braille typing on the touchscreen; the display supports reading and control using a compatible external device. Language, code and connection options are selected in their respective sections. If you do not have a display, skip connection practice. The Braille keyboard may need to be enabled as an input method. Inspect the difference between the two Braille settings. If you use either, locate its language selection. If you have no device, return without pairing; this does not prevent you from continuing training.
Open this chapter in the videoAdvanced settings includes password speech, labels and automatic labelling, single tap, description order, time format, window delay, smart browsing, typing confirmation, long press and physical keyboard shortcuts. Automatic diagnosis checks service and speech paths; its notification step posts a real test notification and shows progress stages. Copy the report or open an email draft. Sending is your choice. Avoid changing developer settings without understanding their effect. Optionally open Automatic diagnosis in Advanced settings. Listen for the real notification and inspect received, accepted, started and completed stages. Review relevant settings if an error appears. Stop the test or leave with Back. Sending the report is not required. Release 114-9 production places Text formatting feedback under Typing in Advanced settings. Choose speech, sound, both, or off, then independently enable font, size, styles such as bold and italic, underline, and foreground or background colours. Use Text formatting information in Reading controls or the three-finger menu for an on-demand report about focused text. If the application does not expose formatting, Reader does not invent exact details; optional screen sampling or the selected AI provider may assist with colour or visual analysis.
Open this chapter in the videoInterface settings adjusts the readability of settings screens. Choose system, light, dark, high-contrast white-on-black, yellow-on-black or sepia themes. Text enlargement combines with the system font size. Bold text, wider line spacing and taller setting rows may help you follow long content. Changes refresh the same screen and are saved. Reset affects only these appearance preferences. Open Interface settings and choose a theme. Read the preview. Enlarge the text and check that long text wraps and controls remain reachable. Optionally reset the appearance, then return to training.
Open this chapter in the videoTraining and help offers the setup wizard, user guide, gesture practice and feedback routes. This interactive guide dedicates a separate step to each settings group. Go to section jumps to any step; the learned switch is your personal marker. First-run training is optional and can be accessed here later. Open Go to section, visit the previous group, then return here. Toggle the learned switch on and off. Leave training and reopen it through Training and help to check that it resumes at the same step. About shows the installed version, developer information, privacy policy, terms, feedback and community links. In an issue report, include your version, exact steps, expected result and observed result. Do not include private messages, passwords or personal screen content. Links may open a browser or email app. Find and read the installed version. You may inspect the feedback email draft, then return without sending. Mark this step if you have finished learning, or revisit earlier sections. This guide uses fourteen video chapters to cover all fifteen settings groups in the application; the last chapter combines Training and help with About. When reporting a problem, include whether you use 114-9 production or 114-10 Beta, Android version, language, screen, steps and observed result. Never include a password, API key or private screen content.
Open this chapter in the videoMenu names, order, visibility conditions and default gesture assignments on this page were checked against the current Android settings definitions and the components that own their behavior.
Verified project state: September 3, 2026, Release 109.
Paths: Settings → Controls → Customize menus is the actual order in the app.
Defaults: Gestures are customizable. The assignments below are reset gesture set 0.
Move one finger around the screen to explore by touch. Accessibility focus moves to the meaningful item under your finger and announces its name, role, state and relevant hint. Swipe right or left for sequential navigation. One-finger double-tap activates the focused item. For split tap, keep one finger on a target and tap with a second finger. Double-tap and hold begins a long press or drag.
Settings → Controls → Customize gesturesThis screen contains Gesture set, Practice gestures, Restore defaults and every assignable gesture below. Logical previous/next directions may be mirrored in right-to-left locales.
On supported hardware, swipe up/down on the sensor selects the previous/next reading control. Left/right are unassigned by default.
Settings → Controls → Customize menus → Customize reading controlsSelect a control with the rotor or three-finger up/down. One-finger up/down then changes the selected value or performs its previous/next action. Options that do not apply to the current screen are omitted. Release 107 remembers the user’s last explicit choice — including Default — across the rotor and three-finger selector. Temporary contextual controls do not overwrite that preference.
The top-left Menu button opens the Settings sections navigation drawer, whose sections are focusable native buttons. Activating a section closes the drawer and opens that section. While the drawer is open, Back closes it first. “Search settings” and voice search search setting names and descriptions; selecting a result opens its section and setting. The source-defined section order follows.
Filter mode (allow all, allowlist or blocklist); allowlist; blocklist; periodic time announcements and interval; quiet-hours switch, start and end. Equal start/end disables quieting.
Enable; online Google/on-device engine; copy to clipboard; source language; target language; download offline packs. Translation runs only when requested from its menu/rotor command.
Gemini Voice Command; Google AI Studio API key; Gemini model and refresh; use on-device Gemma; download/remove the approximately 2.6 GB model; AI system instruction; send a cropped focused-item image. The API key is encrypted with Android Keystore.
Choose recognition language. Map a trigger phrase to an app, system action (Back, Home, Recents, Notifications, Quick Settings, All apps, power dialog) or supported AccessiMind action. Open an existing command to edit/delete it.
Add an app, save the foreground app’s current settings, and manage Saved apps. Each profile reports enabled state and setting count; an enabled profile is applied when that app opens.
AccessiMind Browser: article mode, bookmarks, voice search and on-device AI summaries. App automation: app-specific actions, conditions and bounded repeats.
Braille keyboard opens on-screen Braille training/settings. Braille display opens connected-display, table and connection settings. Availability depends on hardware.
Setup Wizard; Tutorial and user guide; Help and feedback; Contact developer. Practice gestures announces current assignments.
Version/developer; Privacy; Terms; feedback email; Google Group and subscription; Engelsiz Android community.
Correct path: Settings → Advanced settings → Smart automatic labelingConfigure the provider separately at Settings → AI settings → Gemini settings.
The screen shows pending/applied state. Each label offers Apply, Edit and Delete. Management actions import/export only that app’s labels, remove the app, clear all of its labels and grant/revoke cloud consent. Import validates the file and offers replace or keep-current conflict handling.
Analyze current screen contains Label all screen controls and Label focused control. Full-screen mode collects supported unlabeled controls in the visible window only. Focused mode rechecks identity, existing programmatic name and context; it will not apply a result to a stale focus target. Both actions are also available as reading controls. Review results in the app-specific manager.
Search text or list buttons, text and other supported target types on the current screen; choosing a result moves focus to the real item.
Copy focused text or the bounded range between two focus points on the same screen. Password fields are excluded and an unfinished range times out.
The menu tool lists active notifications, filters by source, opens or copies them. Notification management settings instead govern automatic speech; these are distinct.
Check its current summary and gesture set in Customize gestures, then test it in Practice gestures.
Check its switch in Customize context menu or Customize reading controls. Current-app label management also requires that the open app is selected or has stored records.
Check the Gemini key/model or Gemma download/verification, app selection and per-app cloud consent when required.
Email sarperarikan@gmail.com with app/Android version, screen name, exact gesture or menu path, expected result and what was announced. Never send API keys, passwords or private screen content.