Words to Numbers

Paste ‘twenty-three thousand’ and get 23000 — a parser for the numbers people write out in prose.

Checked in a browser on 2026-11-19 · how these tools are tested · runs on your device

What it understands

Hyphenated tens, commas, a stray ‘and’ and the scale words from thousand to quadrillion. It reads left to right, multiplying when it meets hundred and adding when it meets a scale word — the same parse a person does in their head.

Anything that is not a number word stops the parse and the page says so, rather than returning a guess that looks plausible.

    Why this direction is harder

    Speaking a number is mechanical; reading one is ambiguous, because ‘one hundred and one thousand’ could be read two ways. The page follows the common reading and shows the result as digits with separators so a mistake is obvious.

    Currency words are not converted — ‘five dollars’ is not a number, it is an amount.

      Common inputs

      Invoice amounts, legal descriptions and school exercises are the usual sources. The parser is deliberately strict: a word it does not know stops the parse rather than being ignored.

        How Words to Numbers works

        Turn written numbers back into digits: handles hyphens, commas, ‘and’, thousands, millions and more. The method behind Words to Numbers is the single-purpose logic inside wave51.js, and this page shows the result with its assumptions in the open.

        • Open Words to Numbers and provide words to numbers — the page accepts it in the field it shows.
        • For Words to Numbers, text is processed as characters in memory — no services, no grammar engine, no copy kept.
        • The Words to Numbers output appears on the page and can be copied or downloaded in the format it offers; nothing is kept afterwards.

        A worked example for Words to Numbers

        A typical run of Words to Numbers starts with words to numbers and ends with the result shown above; the page reports what it did in plain words so the output can be checked against a known-good case before it matters.

        Run the same check on a paragraph you already counted by hand — the numbers should match exactly, and if they do not, the rule difference is visible in the output. That check applies directly to Words to Numbers with words to numbers.

          When to use Words to Numbers — and when not to

          Reasonable uses: Cleaning up pasted text before it goes into a CMS; Counting words against a limit or deadline; Formatting drafts without opening a heavy editor. Words to Numbers is deliberately narrow: it does one job and states its assumptions instead of hiding them.

          Most people reach for a word processor or an online editor; both add steps and a save dialog. A single-purpose page wins when you need the count, not the document. The honest ceiling for Words to Numbers: for Words to Numbers, what you see is what you typed; the tool does not guess intent.

          • Best case for Words to Numbers: Cleaning up pasted text before it goes into a CMS
          • Also fine with Words to Numbers: Counting words against a limit or deadline
          • Not what Words to Numbers is for: for Words to Numbers, what you see is what you typed; the tool does not guess intent.

          Reading the Words to Numbers result

          Counts follow the rules the page states — Words to Numbers names its own rule so two counters can be reconciled.

            Accuracy and limits of Words to Numbers

            The method is the single-purpose logic inside wave51.js. Words to Numbers prints results at the precision its inputs justify, and the limits below are the honest ones:

            • Words to Numbers: For Words to Numbers, what you see is what you typed; the tool does not guess intent.
            • For Words to Numbers, very large pastes can slow a tab while it renders.

            Privacy and speed on the Words to Numbers page

            The Words to Numbers page is a static file plus wave51.js; there is no upload endpoint behind it, so your input cannot be stored or logged. Closing the tab discards the work.

            Speed follows the build: Words to Numbers loads one stylesheet and one script, with no framework, no web font and no tracker. That is why the tool responds as you type instead of waiting on a round trip.

              What you get from Words to Numbers

              The output of Words to Numbers appears on this page and can be copied or downloaded in the format the page offers. Nothing is queued, nothing is emailed, and there is no usage counter — the text work finishes the moment the numbers are in.

              Because Words to Numbers is built from wave51.js rather than a server call, the practical ceiling is your device: memory for large inputs, and CPU for anything that loops over pixels or pages. On a normal laptop that ceiling sits far above everyday use; on an old phone it arrives sooner, and the page tells you when a job is too big instead of failing silently.

                Questions

                What does Words to Numbers do?

                Turn written numbers back into digits: handles hyphens, commas, ‘and’, thousands, millions and more. Everything runs in your browser; nothing is uploaded.

                How accurate is Words to Numbers?

                The method is the single-purpose logic inside wave51.js. For Words to Numbers, what you see is what you typed; the tool does not guess intent.

                What are the limits of Words to Numbers?

                For Words to Numbers, very large pastes can slow a tab while it renders.

                Is anything uploaded when I use Words to Numbers?

                No. Words to Numbers runs entirely in the browser tab; nothing is uploaded, stored or logged.

                Can I use Words to Numbers offline?

                Yes — Words to Numbers keeps working offline once the page has loaded; the whole tool is a local script.

                Keep going

                More text tools