Mandarin TTS tool

Free Chinese Text to Speech With Mandarin Subtitles Included

Paste simplified or traditional Chinese, pick a Mandarin voice or one of the blends, and download the MP3 with subtitles cut at 16 characters a line. Mixed-in English is read too. No watermark, and no sign-up to listen.

  • 8Mandarin voices
  • 13Mandarin voice blends
  • 5free credits every month
0 bytes / 0 jobs / 0 credits

You can edit or delete these tags directly in your script.

Voice
MeilinChinese · Female
Advanced settings

Output

0.5x1x2x
Output format

Voice mix

The first voice sets pronunciation; the second only lends its character.

Primary voice

Meilin

Sets pronunciation and pacing.

Second voice

A fixed sample at your current mix and speed. No credits used.
  • Commercial use included on the free plan
  • No watermark on any output
  • Preview every voice without signing up

Voices for this page

Mandarin voices, graded honestly

Four female and four male Mandarin voices, all grade D: fine for short clips and announcements, flat over a long paragraph. The blends below borrow a richer timbre; listen to both before you choose.

Browse all voices

What you get

What this Chinese voice generator handles

Two writing systems, characters with several readings, and English scattered through the text. Each one is handled, and the limits are stated.

Simplified and traditional, both read

Paste either script as it is. The engine reads traditional characters directly, so there is nothing to convert before you generate.

One difference worth knowing

Readings for characters with several pronunciations come from a dictionary that knows simplified text best. 银行 comes out as yín háng in simplified and can slip to yín xíng in traditional. If a traditional script has many such characters, convert it to simplified first or fix the readings with the Pronounce as tag.

Listen to every Mandarin voice

Multi-reading characters caught

On a paid plan, click Tidy up script: the AI pass checks 行, 长, 还, 重, 得 and forty-odd others against their context and tags the intended reading.

How the fix works

The engine always takes the most frequent reading, so 一行 in a table is read xíng unless told otherwise. The tidy pass swaps in a stand-in character with the right sound for the voice only; your original text stays on screen and in the subtitles. You can add the same Pronounce as tag by hand for names and place names.

Need Japanese too? See the Japanese page

English inside Chinese text

Brand names, app menus, and English terms in a Chinese script are read in place by the Mandarin voice, with an accent, rather than skipped.

How mixed scripts are rendered

The script is split by writing system, the Chinese runs go through the Mandarin reader and the Latin runs through the English reader, and both are joined into one take by the same voice. Numbers, currencies, and units are spelled out in Chinese by the tidy pass, so 40,000 is read as 四万.

Reading mostly English? Use the main generator

Subtitles

Chinese subtitles that fit a phone screen

Chinese captions need short lines: 16 characters is the usual width for a vertical video. The SRT here is cut for that, straight from the same render as the audio.

  • SRT and WebVTT with every job, on the free plan included
  • Lines capped at 16 characters, with breaks at Chinese punctuation
  • Cues anchored to real pauses in the audio, aligned per sentence
  • Edit a sentence, render again, and the cues move with the audio

The honest limit: Mandarin voices report timing per sentence, not per word. The worker detects the pauses in the finished audio and anchors each cue to them, which is accurate for ordinary captions but not for karaoke-style highlighting. Word-level timing is available on the English voices.

Generate with subtitles
1
00:00:00,160 --> 00:00:03,080
今天介绍一个新功能。

2
00:00:03,300 --> 00:00:06,520
先打开设置页面，找到右上角。

3
00:00:06,760 --> 00:00:09,400
点一下，字幕就生成好了。

Voice blends

Mandarin blends that borrow a richer timbre

Each blend keeps a Mandarin voice in charge of pronunciation and mixes in a higher-graded voice for character: Delaney's husk, Maya's clarity, a British lesson tone. The accent that comes with it is part of the appeal.

Magnetic Mandarin

Mandarin female with a throaty edge.

Meilin 50% + Delaney 50%

Husky announcer

Mandarin announcements with a throaty edge.

Ruoxi 20% + Delaney 80%

British-toned Mandarin

Mandarin pronunciation, British lesson timbre.

Ruoxi 10% + Freya 90%

Clear Mandarin

Mandarin with Maya's clarity.

Ruoxi 20% + Maya 80%

Instructional Mandarin

Mandarin with a British instructional tone.

Ruoxi 10% + Nell 90%

Soft Mandarin

Two Mandarin voices blended, gentler than either alone.

Ruoxi 20% + Tongtong 80%

Mandarin, Japanese tone

Mandarin pronunciation with a Japanese timbre.

Ruoxi 10% + Yukiko 90%

Magnetic Mandarin male

Mandarin male with a throaty edge.

Weilun 50% + Delaney 50%

Magnetic Mandarin male · lighter

Mandarin male, throaty edge held back.

Weilun 70% + Delaney 30%

Clear Mandarin male

Mandarin male with Maya's clarity.

Weilun 50% + Maya 50%

Mandarin course intro

Lesson opening in Mandarin.

Weilun 30% + Nora 70%

Mandarin report

Report-summary read, two male voices.

Zhiyuan 30% + Marcus 70%

How voice blending works

A blend puts a Mandarin voice in charge of every character's pronunciation and pacing, then mixes in the second voice's timbre at the percentage you choose. Because the second voice may come from any language, these blends sound fuller than any single Mandarin voice here while the Chinese stays correct. Preview a blend at no cost, then save the ones you like in this browser.

Try a blend in the generatorHear the single voices first

How it works

How to turn Chinese text to audio in three steps

  1. A calligraphy brush resting on an ink stone, with sound waves rising from it.STEP 01

    Paste the Chinese text

    Simplified or traditional, with any English left in place. The generator shows the byte count and the exact credit cost before you submit.

  2. A bamboo scroll unrolling beside sound waves, for choosing a Mandarin voice.STEP 02

    Pick a Mandarin voice or a blend, then tidy

    Play the samples, choose a voice, and run Tidy up script so multi-reading characters and numbers are read the way you meant.

  3. A monitor showing two lines of text, for the subtitle download.STEP 03

    Download the MP3 and the subtitles

    The audio and the SRT and VTT files appear in My Speech. Import them into your editor together and the captions land on each sentence.

Free plan, plainly

Free Mandarin TTS, and what a paid plan adds

Everything on this page works on the free plan. Paying buys queue priority, WAV, and longer file retention.

CapabilityFreeStarterCreator
PriceFree$6.99 a month, or $69.99 a year$15.99 a month, or $159.99 a year
Credits per month5 credits100 credits300 credits
Standard English characters per job10,000 for 1 credit10,000 for 1 credit10,000 for 1 credit
QueueFree queue, no promised timeSkips the free queueSkips the free queue
Output formatMP3MP3 and WAVMP3 and WAV
SubtitlesSRT and VTTSRT and VTTSRT and VTT
Script tidy (AI)Not availableIncludedIncluded
Premium voicesNot availableComing soonComing soon
Credit packsNot available100 credits for $5100 credits for $5
File retention72 hours30 days30 days
Commercial useIncludedIncludedIncluded
WatermarkNoneNoneNone
How credits are counted

One credit runs one job, and that job comes with room for 10,000 UTF-8 bytes: about 10,000 characters of English, or about 3,300 Chinese or Japanese characters. That is a whole article or a long script in a single go. Longer pieces are never turned away either, since each further block of 10,000 bytes adds a credit and what comes back is still one continuous file. Credits are reserved when you submit and refunded in full if a job fails.

Compare the plans

FAQ

Questions about Chinese TTS, answered

  • Does it support traditional Chinese?

    Yes. Traditional and simplified characters are both supported natively, with no conversion step. Characters with several readings are resolved more reliably in simplified text, so convert a long traditional script first if readings matter.

  • How does it handle characters with multiple pronunciations?

    The engine picks the most frequent reading. Tidy up script scans about fifty common multi-reading characters, 行, 长, 还, 重 and so on, and tags the intended reading from context. Add the Pronounce as tag yourself for anything it misses.

  • Is Cantonese available?

    Not yet. Every voice on this page speaks Mandarin. Cantonese is on the roadmap without a date, so do not plan a project around it.

  • Why does Chinese text use credits faster than English?

    Standard billing counts UTF-8 bytes, and each Chinese character is three bytes, so one credit runs one job with room for about 3,300 Chinese characters. A longer script adds a credit per extra block. The generator shows the exact credit cost before you click generate.

  • Can I get Chinese subtitles with correct timing?

    Yes, with sentence-level alignment. Mandarin voices do not report per-word timing, so each cue is anchored to a real pause in the audio and cut at 16 characters a line. That is right for captions, not for word-by-word highlighting.

  • Are the voices male or female?

    Both: four female and four male Mandarin voices. All are grade D on our honest scale, meaning clear but flat over long passages, which is exactly why the blends exist.

  • Can I mix Chinese and English in one text?

    Yes. Keep Chinese as the main script and the English words are read in place by the Mandarin voice, with a Chinese accent. If the script is mostly English, choose an English voice and keep Chinese to a minimum, since English voices cannot read Chinese characters.

Generate your Mandarin audio

Paste the text, pick a voice or a blend, and download the MP3 with Chinese subtitles.

Open the generatorView pricing