Free Online Text to Speech, Nothing to Install

What "free online text to speech" usually leaves out

Most tools that rank for the phrase are free in one of three ways: free for a few hundred characters, free with the voices you would actually want locked behind a plan, or free until you try to download. It is worth being precise about which kind this is. All forty voices are available on the Free plan. Listening and regenerating cost nothing. The balance — roughly five minutes of finished English audio, more in Chinese, Japanese and Korean — is spent only when you download, and it is a one-time balance rather than a monthly one. The Free plan also caps generation at 30 lines a day and is licensed for personal, non-commercial use.

Paid plans change the numbers, not the voices: a monthly balance in place of the one-time one, more lines a day, commercial use, and on Plus and Pro the per-line pace control and subtitles. Nothing about the sound of the voices changes when you pay.

Getting a natural read out of any text to speech

Text written for the eye contains things that do not survive being spoken: long parenthetical asides, three nested clauses, a paragraph that relies on the reader glancing back. Ten minutes of adaptation — splitting sentences, cutting an aside, turning a list into a line per item — does more for the result than any voice setting. This is not a workaround for a limitation; it is the same edit a human narrator would ask for before a session.

Two habits matter most. Punctuate deliberately, because punctuation is the only timing control you have on the Free plan and the main one on any plan. And keep one idea per line: a sentence that swings from good news to bad is rendered as an average of both, and splitting it is the fix.

Numbers, dates and acronyms follow the conventions of the project's language and are usually right; text that mixes conventions is where mistakes appear. Audition the lines with figures in them first, and spell out anything whose reading matters.

When this is the right tool, and when it is not

If you want a web page or a PDF read aloud while you do something else, a reader extension is a better fit — it follows the text and does not ask you to paste. If you need a single voice reading an hour of prose with no direction, almost any online text to speech will do it, and some are cheaper per hour.

CastDub earns its place the moment the text has more than one voice in it, or the moment a line needs to be said a particular way. A dialogue, a presenter with a customer, a story with a narrator and characters, an explainer where the call to action should land differently from the paragraph before it: these are the cases where one flat voice per file stops being enough, and where casting and per-line direction — which is what this text to speech is built around — start to matter.