Number to Words Converter

Spelling a number out in words by hand is a small, repetitive task that shows up in a surprisingly wide range of contexts: writing a check (most banks still require or accept the amount spelled out in words as a fraud-prevention redundancy alongside the numeral), drafting a legal or financial document where numeric amounts are conventionally written out in full, writing a cheque number into an invoice template, or simply double-checking that you have read a large number correctly (is `1,000,000,000` a billion or a milliard? in short-scale English, always a billion). This tool automates that conversion for any number you type, live, as you type it. English has two historical systems for naming large numbers — short scale (used in the US and, since the late 20th century, standard in the UK and most English-speaking countries: a billion is 10⁹, a trillion is 10¹²) and long scale (used in some European countries: a billion is 10¹²). This tool uses short-scale naming exclusively, since it is now the standard convention for English, and supports the full sequence thousand → million → billion → trillion → quadrillion → quintillion, comfortably covering the 15-digit integers explicitly required for a tool like this and well beyond — a plain 15-digit whole number sits solidly in the trillions, and this tool's scale table reaches two full steps past that. The conversion works by treating the number as a plain digit string rather than converting it through JavaScript's native floating-point `Number` type for the actual word-building logic. This avoids a subtle but real class of bug: binary floating-point numbers cannot represent every decimal value exactly (the classic example being `0.1 + 0.2` not equaling exactly `0.3` in JavaScript, or any language using IEEE 754 floats), so an approach that parses your input into a `Number` and then tries to re-extract exact decimal digits from it can silently produce the wrong digits for certain values. Instead, the input string is validated with a regular expression, split on the sign, the decimal point, and then processed as plain digit groups: the integer part is chunked into groups of three digits from the right (thousands, millions, billions, and so on), each group of up to three digits is spelled out on its own ("one hundred twenty-three"), and the appropriate scale word is appended ("million", "billion", etc.) based on that group's position — exactly mirroring how you would read the number aloud yourself, one comma-separated group at a time. Decimals are handled two different ways depending on the mode you choose, because "the right way to say a decimal" genuinely depends on what the number represents. In standard mode, each digit after the decimal point is spoken individually — `3.14` becomes "three point one four", not "three point fourteen" — which is the mathematically unambiguous convention (saying "fourteen" for `.14` could be misread as `.140` versus `.14` versus a totally different value depending on context, whereas reading each digit removes all ambiguity, which is exactly how numbers are conventionally read aloud in scientific and everyday contexts alike, e.g. reading out a phone number or a decimal measurement). Currency mode instead renders exactly two decimal digits as a "cents" fraction in the `and NN/100` format used on checks and financial documents — `123.45` becomes "one hundred twenty-three dollars and 45/100" — truncating (not rounding) to the first two decimal digits if more were entered, and correctly singularizing to "dollar" (not "dollars") when the whole-number part is exactly 1. Negative numbers get a straightforward "negative" prefix (`-42.5` → "negative forty-two point five"), and zero is handled as an explicit special case rather than falling out of the general algorithm by accident, since the general group-by-group logic would otherwise produce an empty string for zero (there is no group to spell out when every digit is zero) — the tool checks for this and returns the word "zero" directly.