A guided setup walks you through enabling and selecting the keyboard with steps that fully cooperate with TalkBack, and confirms each step (keyboard enabled, keyboard chosen) with a spoken toast message.
You can choose from more than 200 keyboard layouts covering more than 30 languages, including multiple variants per language, so you can pick the exact layout that matches your physical keyboard and regional preferences.
Speak instead of typing, with a choice of recognition: an on-device local model that works offline and supports many languages, or cloud recognition through OpenAI or Google Gemini once you enter your own API key. The classic system voice input is always available as well, so you can pick whichever option suits you best. Voice input can be started via the microphone key on the keyboard or through the keyboard’s Quick Menu.
You can choose how taps register on the keyboard. In “Lift to Type” you slide your finger across the keys, hear each one as you touch it, and only commit when you lift; in “Double Tap to Type” the first tap selects and reads the key, and a second tap confirms it — both modes give you a chance to verify a key before sending it.
When you hover a key while Shift, Alt, or Ctrl is active, the keyboard reads the character that will actually be typed ( e.g., over “1” with Shift it reads “Exclamation mark”) instead of the unmodified label.
When you hover a layout-switch key, the name of the sub-layout that key would take you to is read aloud. You can decide whether to switch to letters, numbers, symbols, or emojis without committing first.
When enabled, a list of recently copied text snippets is shown above the keys. Tap any entry to insert it instantly into the current text field, so you can reuse something you copied earlier.
Holding down the space bar opens the Quick Menu, giving you quick access to several features such as switching between your selected keyboard layouts and starting voice input.
Available only when TalkBack is turned off, a quick two-finger swipe on the keyboard triggers a shortcut: swipe up to start voice input, swipe down to open the Quick Menu, and swipe left or right to switch to the previous or next keyboard layout — fast access to common actions without hunting for the corresponding key.
The keys you explore and the actions you trigger are read aloud, so you can use the keyboard without looking at the screen. You can turn the speech on or off, set how fast it speaks, choose a comfortable pitch, and have upper- and lower-case letters distinguished by a varying pitch.
Letters can be read using the NATO alphabet (“Alpha”, “Bravo”, “Charlie”, …), so similar-sounding letters like “B” and ” D” can never be confused.
Descriptive names are spoken for punctuation, symbols, and emojis (such as “Open parenthesis”, “Euro”, or “Question mark”), so you can find them without seeing the glyph.
Important changes are spoken as they happen: when the keyboard appears or disappears, when you switch between letters, numbers, symbols, or emojis, and whenever Shift, Alt, or Ctrl change between inactive, active, and locked. You always know which keyboard, sub-layout, and modifier state is currently in effect.
A short click plays each time a key is typed, with several sound styles (Typewriter, Mini, Soft) and adjustable volume, giving you a quick audible confirmation alongside the speech.
A short vibration is triggered when a character is typed and when a swipe gesture (such as moving the cursor or selecting text) is recognized. Each can be enabled separately, so you get physical confirmation of input without needing to see the screen.
The entire settings screen is comfortable to use with TalkBack: switches and their labels are grouped as a single element, section headings are announced with a “Section:” prefix, and numeric settings such as keyboard height or speech rate use explicit “+” and “−” buttons (which announce when the minimum or maximum is reached) instead of sliders.
When you finish a word with a separator like space or punctuation, the keyboard checks the word against the dictionary for your active layout’s language and quietly replaces it with the correct spelling if a match is found — for example, typing “teh” becomes “the”. The original word is restored by pressing backspace once, so an unwanted correction is easy to undo. An optional confirmation sound can be enabled and its volume adjusted, and the corrected word can also be spoken aloud when speech output is on. Auto-Correct stays out of the way in password fields, where no replacement is ever performed.
You can define your own short triggers in the options that expand into longer text — for example, “brt” can expand into “Best regards, Tom” or “eml” to “me@mydomain.com”. As with Auto-Correct, the expansion happens when you end the trigger with a separator such as space or punctuation, and a single backspace restores the original trigger if you didn’t want the expansion. Text Shortcuts take priority over Auto-Correct, so a shortcut you defined is never accidentally “corrected” away. An optional confirmation sound with its own volume can be enabled, and the expansion can be announced through speech output as well.
When you hold the Delete or Backspace key, the keyboard pauses briefly before it starts repeating, then removes characters at a steady pace. You can adjust both the initial delay and the repeat speed, so you can delete a single character without overshoot or clear a long passage quickly.
Everything that happens while you use the keyboard — layout changes, settings, and device details — is recorded, and you can clear, export, or e-mail the log, which helps when reporting an issue or asking for support.