Is there a keyboard based on sound signatures?
The keyboard in touch devices encroaches visual space. Nowadays, on iPad Pro I use the floating keyboard and swipe gestures.
I wish there is a shorthand style sound signatures as keys, swiped and connected to build words. And a layout that optimizes the most common words and sentences, computed using machine learning. The shorthand techniques also have logograms to make transcription easier.
The word thought is 7 letters, but can be represented by three sounds "th"-"o"-"t". Probably "th" and "t" have to be kept together for a better mental model. And it would be a simple swipe gesture, instead of the complex one based on actual spelling and QWERTY keyboard layout.
Do you know any such keyboard exists?
Comments
Each character in korean is a syllable with an (optional) starting and ending consonants and a vowel in-between. Each stroke (approximately) matches each consonant and vowel. Typing korean inputs these 'strokes' that build each syllable and a sequence of them make a word.
So any korean keyboard would fit you definition though not in english. With the written language being strongly phonetic, anything can be sounded/written out. English consonants don't map 1:1 so there's inaccuracy/ambiguity.
Consonants/vowels isn't very high density though. It would be better to have a sort-of Huffman coding that a human can synthesize/parse based on frequencies and context. The world of shorthand writing probably has such examples.