How words are turned into numbers
A plain-English look at how spelled-out numbers are parsed, why scale words like hundred and thousand matter, and where the logic stops.
Reading number words left to right
English number words are not just a list of digits, they are a tiny grammar. The parser walks the words from left to right and keeps a running total for the current group, adding units such as 'seven' and tens such as 'sixty' as it meets them. So 'sixty seven' becomes sixty plus seven, giving 67. Because the values simply accumulate, the order inside a group is forgiving, and stray spacing or capitalization does not change the outcome.
Why hundred, thousand and million behave differently
Scale words are multipliers, not additions. When the reader hits 'hundred' it multiplies whatever it has gathered so far by 100, so 'three hundred' is 3 times 100. 'Thousand' and 'million' are bigger multipliers that also bank the running group into the grand total and reset the counter, which is how 'two thousand five hundred' correctly lands on 2500 rather than muddling the pieces. If a scale word appears with nothing before it, it is treated as one, so a lone 'hundred' is 100.
The filler words and punctuation it ignores
Real writing is messy, so the parser strips out the noise before it counts. Hyphens are swapped for spaces, which is why 'ninety-nine' and 'ninety nine' behave the same. The connective 'and', as in 'one hundred and five', is removed because it carries no numeric value in this context. Mixed case is flattened and repeated spaces collapse, so text copied from a document usually converts without any cleanup on your part.
The limits worth knowing
This converter handles whole cardinal numbers up to the millions and nothing beyond that vocabulary. It does not know 'billion', ordinal forms like 'twenty-first', fractions, decimals or negative values, and any unknown token stops the parse with a clear error. It also adds cardinal values literally, so a spoken year like 'nineteen eighty four' becomes 103 rather than 1984. Knowing these boundaries lets you phrase input in a way the tool understands.