# Text to Speech Online - Free TTS Reader with Natural Voices

> Free text to speech: near-human AI voices run in your browser, unlimited, downloadable as WAV, in English and Chinese. System voices cover many more languages.

- URL: https://qilingtools.com/en/text-to-speech
- Category: Productivity (https://qilingtools.com/en/office)
- Tags: [Audio](https://qilingtools.com/en/tags/audio), [Text](https://qilingtools.com/en/tags/text), [Productivity](https://qilingtools.com/en/tags/office)
- Last updated: 2026-10-26
- Price: Free, no sign-up
- Privacy: Runs locally in the browser; files are never uploaded

Turn text into speech: paste an article, lesson, script or foreign-language sentence and hear it read aloud. There are two modes, both free and both running in your own browser. "High-quality AI voices" uses Kokoro, an open-source neural text-to-speech model (Apache-2.0), to generate speech right on your device: near-human audio, no server involved, no length limit, and once it has read your text you can download the whole thing as a WAV file — handy for video narration, audiobooks, listening practice and course material. English and Chinese each have their own model: 28 English voices (20 American, 8 British) and 100 Chinese voices (45 male, 55 female), and the Chinese model reads numbers and dates the Chinese way and handles English words mixed into Chinese text. The first use of a language downloads its model (about 102 MB for English, 116 MB for Chinese), once — after that they stay in your browser, so it opens instantly and works offline. You see the size before anything downloads, with an extra warning on mobile data. Synthesis uses several processor cores in parallel and plays while it generates, so a long article starts after a few seconds and then never stalls. "System voices" uses the voices built into your computer or phone (Web Speech API): nothing to download and many languages including Chinese and Japanese, but playback only, no download. Both modes let you adjust speed and volume, pause and resume, and highlight the sentence being read — great for shadowing and proofreading. Ten voice personas are built in — cute shota boy, hot-blooded teen, tsundere loli, soft sweet girl, commanding mature woman, warm intellectual woman, cold CEO male lead, mellow uncle / elder storyteller, crisp tech female anchor and steady science narrator. In AI mode a persona lists the few voices whose measured pitch fits it best; the English voices are all adult voices, so child personas are available only in Chinese and with system voices. Voices and personas are searchable.

## How to use

1. Pick "High-quality AI voices" or "System voices", then paste the text you want read aloud
2. Choose a voice persona or a voice (in AI mode, press Preview first), and adjust speed and volume
3. Press Play, then pause, resume or stop whenever you like; in AI mode, download the WAV when it finishes

## Features

- High-quality AI voices: the open-source Kokoro model runs on your device and sounds near-human
- No server and no length limit — your text never leaves your machine
- Download the result as WAV in AI mode
- 28 English voices and 100 Chinese voices, each with a preview
- The Chinese model also reads English words mixed into Chinese text
- Model downloads once and then works offline; size shown first, extra warning on mobile data
- Parallel multi-core synthesis that plays while it generates
- System voices cover every language your browser provides: English, Chinese, Japanese and more
- 10 voice personas: boy / loli / mature female / CEO / uncle / news
- Searchable voices, personas and languages
- Speed and volume controls; play, pause, resume and stop
- Sentence highlighting while reading

## FAQ

### Can I download the speech as an MP3?

In "High-quality AI voices" mode you can download the whole audio after one read-through. It is a WAV file, which virtually every editor and player accepts; convert it with any audio converter if you need MP3. "System voices" cannot be downloaded: the browser's Web Speech API only plays audio and never gives the page the audio data.

### Why does AI mode download 100+ MB first?

The whole neural network runs inside your browser instead of on a server, so the model has to be on your device: about 102 MB for English and 116 MB for Chinese, downloaded only for the language you use. It downloads once and stays in your browser's cache — next time it opens instantly, even offline. The size is shown before anything starts, you can cancel at any time, and you get an extra warning on mobile data or data saver.

### Do the AI voices speak Chinese or other languages?

English and Chinese. Pick the language at the top of AI mode: Chinese has 100 voices and also reads English words mixed into Chinese text (with a slight Chinese accent). For other languages use "System voices": Microsoft Edge has many natural online voices, Windows can add voices under Settings › Time & language › Speech, and macOS/iOS let you download high-quality voices under Accessibility › Spoken Content › Voices.

### Synthesis is slow or choppy. What can I do?

AI voices are computed entirely on your processor, using several parallel workers sized to your core count. A recent computer runs about three times faster than real time, so playback is continuous after a few seconds; on older computers or phones there may be short pauses between sentences — in that case let it finish once and listen to the downloaded WAV.

### Are the voice personas real voice changing?

No — and not voice cloning either. In AI mode a persona picks the existing voices whose measured pitch is closest and presets the speed; all English voices are adult voices, so child personas exist only for Chinese. With system voices a persona picks the closest system voice and applies preset pitch and speed, so the result depends on which voices are installed.

### Is my text uploaded? Is it free?

AI mode synthesizes everything locally; your text never leaves your machine. System voices don't upload text either, but a voice marked "online" is processed by the browser vendor's speech service, while voices marked "local" work fully offline. Both modes are free and unlimited, with no sign-up.

## Related tools

- [QiLing Mind](https://qilingtools.com/en/mindmap): QiLing Mind is a free online mind map editor: add nodes with Tab/Enter, drag to restructure, import and export XMind and Markdown, export PNG/SVG/PDF.
- [Password Generator](https://qilingtools.com/en/password): Free online random password generator: choose length, character sets and look-alike exclusions, generate several at once with a strength meter, all locally.
- [QR Code Generator](https://qilingtools.com/en/qrcode): Free online QR code generator: turn URLs, text, Wi-Fi or contact cards into QR codes with custom size, colour and error correction. Crisp PNG or vector SVG.
- [Word Counter](https://qilingtools.com/en/word-count): Free online word counter: live counts of characters with and without spaces, words, CJK characters, lines, paragraphs, sentences, bytes and reading time.
- [Case Converter](https://qilingtools.com/en/case-convert): Free online case and naming converter: get UPPERCASE, lowercase, Title Case, camelCase, PascalCase, snake_case, kebab-case and CONSTANT_CASE at once.
- [Lorem Ipsum Generator](https://qilingtools.com/en/lorem): Free online Lorem Ipsum generator: classic Latin placeholder text or Chinese filler by paragraph, sentence or word, with one-click copy for mockups.
- [Sort & Dedupe Text](https://qilingtools.com/en/text-sort): Free online text tool: sort lines alphabetically or by length, remove duplicate and empty lines, trim whitespace, optionally ignoring case. Runs locally.
- [Sign & Stamp PDF](https://qilingtools.com/en/pdf/sign): Free online PDF signing tool: drop in a handwritten signature or company stamp, drag it into place on a page preview, adjust size and opacity, then export.