How to Convert Words to Number: Reading Spelled-Out Amounts Back Into Digits

Most number tools go one way: you feed them 1234 and they hand you "one thousand two hundred thirty-four." The reverse trip is the one that actually shows up in real work. A contract spells the amount out. A cheque writes it twice — once in a box, once on a line. A meeting transcript says "about three hundred fifty units" instead of 350. To get those into a spreadsheet, you have to turn the words back into figures, and that is where most people slow down and miscount a zero.

This guide explains how spelled-out numbers parse back into digits, walks through a worked example, and covers the parts that trip people up: scale words, hyphenated tens, the British "and," and decimals. If you just want the answer, the Words to Number Converter reads any phrase the moment you type it.

How number words actually combine

English number words follow two operations, and once you see them you can parse anything by hand. Small words add within a scale, and scale words multiply.

Take "three hundred forty two." The parser keeps a running bucket. It reads "three" (3), then hits "hundred" and multiplies what it is holding by 100, giving 300. Then "forty" adds 40, and "two" adds 2. The bucket is now 342. Every small word between scale words is simply summed; every scale word is a multiplier on whatever came before it.

That is the whole grammar. "Twenty-three" is 20 + 3. "Six hundred" is 6 × 100. "Forty-two thousand" is (40 + 2) × 1000. The hyphen in "twenty-three" is just typography — it joins a ten and a unit that were always going to add together, so the parser strips it and treats the two words as one bucket entry.

The only subtlety is the scale stack. Hundred is a small multiplier; thousand, million, and billion are big ones that "flush" the bucket upward and lock it in. When the parser meets "thousand," it takes everything accumulated so far, multiplies by 1000, adds it to a grand total, and resets the bucket for the next group. That reset is why "twelve thousand three hundred" reads cleanly as 12 × 1000 + 3 × 100 = 12300 rather than collapsing into one confused sum.

A worked example: "one thousand two hundred thirty-four"

Let me trace this one all the way through, because it touches every rule at once.

"one" → bucket = 1
"thousand" → flush: total = 1 × 1000 = 1000, bucket resets to 0
"two" → bucket = 2
"hundred" → bucket = 2 × 100 = 200
"thirty" → bucket = 200 + 30 = 230
"four" → bucket = 230 + 4 = 234
end of phrase → total = 1000 + 234 = 1234

The hyphen in "thirty-four" never matters; it adds the same way "thirty four" would. And notice the bucket got flushed exactly once, at "thousand," which is what keeps the thousands group and the hundreds group from bleeding into each other. The first time I traced this on paper I expected the parser to need lookahead — to peek at the next word before deciding. It doesn't. Add small words, multiply on scale words, flush on big scales, and the answer falls out left to right in a single pass.

Parsing amounts from cheques and contracts

This is where reading words back into digits earns its keep. Legal and financial documents deliberately write a number twice — the words are the authority, the figures are the convenience — so the two should always agree. When they don't, someone has a problem, and you want to be the one who caught it.

A cheque that reads "two thousand and five" should match a figure box of 2005. Paste the words, read 2005, compare. If the box says 2050 instead, you have spotted a transposed zero before it cleared. A contract clause reading "the sum of twelve thousand three hundred dollars" needs to land in your ledger as exactly 12300 — not 12,030 because you lost count of place values while reading. The machine never loses count.

Two things from finance documents that the Words to Number Converter handles without you thinking about them:

The British "and." "Three hundred and seven" reads 307; the "and" is glue and gets ignored. American style writes "three hundred seven" and lands on the same 307. You never have to pick a dialect.
Stray commas and mixed case. "Two Thousand, and Five" parses the same as "two thousand and five." Documents are messy; the parser tolerates the mess.

Scales, hyphenated tens, and the edge cases

A few patterns cause more confusion than they should, so it is worth naming them.

Hyphenated tens are the most common. Compound numbers from twenty-one to ninety-nine are conventionally hyphenated: forty-two, eighty-seven. The hyphen carries zero arithmetic meaning — it is there for readability — so "forty-two" and "forty two" are identical to the parser. Both are 40 + 2 = 42.

Scale words must climb, not repeat. A well-formed number uses each big scale once and in descending order: billion, then million, then thousand, then the bare hundreds. "Two million three hundred thousand" is fine; "three thousand two million" is not a number anyone writes. If you feed a parser something genuinely malformed, the safe behavior is to return empty rather than invent a value — guessing turns a typo into silent corruption.

Decimals read digit by digit. After the word "point," each following word must be a single digit zero through nine. "One point five" is 1.5 and "three point one four" is 3.14. What you cannot do is write "one point fifteen" — the fraction "fifteen" is two digits and the rule only reads one at a time. Spell it as "one point one five" for 1.15.

The sign goes first. Negative numbers start with "negative" or "minus": "negative forty-two" is -42. Tucking the sign at the end ("forty-two negative") will not parse, because the sign word has to lead the phrase before any digits arrive.

When you need the trip in the other direction

Reading words into digits is only half the round trip. If your job is the reverse — turning 1234 into "one thousand two hundred thirty-four" for a contract draft, an accessibility label, or a spelled-out cheque amount — you want a generator instead of a parser. The Number to Words Converter does exactly that, and the two tools pair naturally: write a figure out, then paste the words back to confirm they round-trip to the same number. If they don't, your spelling has a bug, and catching it on a draft beats catching it on a signed page.

For everyday re-keying — a transcript quantity here, a cheque amount there — the parser is the faster half of the pair. Type the words, read the digits, copy, move on. No counting places in your head, no miscounted zero, no place-value puzzle. The grammar is just add-and-multiply, and once a tool runs it for you, spelled-out numbers stop being something you have to slow down and decode.

Made by Toolora · Updated 2026-06-13