How words are counted
Splitting on whitespace is the obvious approach and it is wrong for a large share of the world's text. It counts "state-of-the-art" as one word, and it counts an entire Chinese or Japanese sentence as one word because those scripts do not put spaces between words.
This counter uses Intl.Segmenter, the browser's built-in Unicode segmentation, which understands word boundaries per script. On older browsers without it, the tool falls back to whitespace splitting — accurate for space-separated languages, approximate elsewhere.
Characters with and without spaces
Both are shown because different limits mean different things. Twitter and SMS count everything; some academic and legal limits exclude spaces. Note that both counts here treat an emoji as one character, which is what a person means — though some systems count the underlying code units instead and will see a single emoji as two or more.
Reading and speaking time
Reading time assumes 220 words per minute, the usual figure for adults reading prose on screen. Speaking time assumes 130, which is closer to a measured presentation pace than to conversation. Both are estimates: dense technical text reads slower, and a script read aloud with pauses runs longer than the raw count suggests.
Where sentence counting is approximate
Sentences are detected by terminal punctuation followed by whitespace. That mis-splits on abbreviations — "Dr. Smith" and "e.g. this" both register as sentence ends — and no regular expression solves it properly, because the same character genuinely does both jobs. Treat the sentence count as close rather than exact.