The trouble with figures shows up the second you dictate one. You say “twenty three dollars and fifty cents” and your Mac writes it out as words. You say a year and get a random mix of digits and spelled-out numbers. Prices, dates, times, percentages, and phone numbers all come out looking like a transcript of someone reading aloud, not text you can actually use. So you stop, grab the mouse, and fix it by hand, which defeats the point of dictating.
Why built-in dictation mangles figures
Apple Dictation auto-formats some numbers, but not consistently. It spells small numbers out as words, and it has no automatic handling for currency, dates, or times. The workaround of saying “numeral” before a number is fiddly and easy to forget, and even then you have to chain spoken symbol commands (“dollar sign”, “point”) just to get $23.50, with no clean path to a date like May 1, 2026. So you memorize spoken tricks, or you clean up afterward, every single time. If your dictation numbers are not working the way you expect, this is usually why.
How to dictate numbers on Mac, no commands needed
With Dictato, you speak naturally and the digits and symbols come out clean. No setup, no toggle, nothing to memorize. A few examples:
- Say “twenty three dollars and fifty cents” and get $23.50
- Say “two point five billion dollars” and get $2.5 billion
- Say “fifty percent” and get 50%
- Say “may one twenty twenty six” and get May 1, 2026
- Say “quarter past two” and get 2:15
- Say “twelve thirty am” and get 12:30 am
- Say “seventy two degrees fahrenheit” and get 72 °F
- Say “two thirds” and get 2/3
- Say a phone number and get tidy grouping like 123-555-0142
It follows the conventions of the language you speak and your regional date order, so US dictation gives “May 1, 2026” while a day-first region gives “1 May 2026”. It also cleans up forms you have already written, turning “May 1st” into “May 1” and “12:30 PM” into “12:30 pm”.
Why it stays reliable
This formatting is rule-based and runs entirely on your Mac in a tiny fraction of a second. The same input always produces the same output. It cannot invent a digit or drop a word, because it is not guessing. You learn once what clean output looks like, and you get it every time.
It is also conservative by design. It only acts on clear number phrases, so ordinary sentences are left alone. “Three cats” stays “three cats”, and “I may go” stays “I may go”. You get clean figures without your prose being reshaped into something you never said.
One more thing matters here: the number, date, and currency formatting always runs last. So even when other cleanup happens, nothing circles back and re-mangles a figure you dictated. The number you said is the number you keep.
The honest payoff
You will still proofread, because everyone should. But you will stop hand-fixing prices, dates, and times after every single dictation. Dictato handles the formatting locally, with nothing sent to the cloud, no account to create, and no subscription. You pay once and use it for life. For anyone who dictates figures often, that is the gap between voice-to-text that fights you and voice-to-text that just works.