About ReadAnyText

Why it exists

Most free text-to-speech tools stop at a play button. If you are making a video, narrating a document or building something accessible, you usually need more than that - the timings, more than one voice, control over where the delivery pauses. Those things are not hard to provide, so this site provides them and does not charge for them.

What it does

• 322 neural voices across 75 languages, each with its own page where you can try it.
• Word-accurate subtitles. Every render produces SRT and WebVTT files timed to the individual word, not guessed from the text.
• More than one speaker. Label lines with a name and cast a different voice to each; everything is stitched into one file.
• Pauses where you want them. Drop in [pause 1s] and the audio breathes there.
• PDFs. Drop one in and it is read in your browser the file is never uploaded.

How it is paid for

The free plan is the whole studio: every voice, every export, 7,000 characters per render and a two-minute wait once that is used up. Paid plans raise that ceiling so a long article or a chapter fits in one render, and keep listen links up for longer. They are what pays for the free one. The pricing page has the figures.

The voices

Speech is synthesized by a third-party neural voice service.

Who makes it

ReadAnyText is built and maintained by Tharusha Kulasinghe. Feedback, bug reports and voice requests are welcome through the contact page.