Chinese text to speech with native Mandarin voices
Stacy, Amy, Jason and James are Mandarin speakers - tones, tone sandhi and the characters that have two readings. Paste your text, hear it free.
Try it free →150 free credits on signup · no subscription, pay per character
Hear the difference
The same Chinese sentence, twice. Stacy is a Mandarin speaker; Aria reads the characters with English intonation, which flattens the tones - and in Chinese the tone is not decoration, it is part of the word.
The Chinese voices
All 4 run on the ElevenLabs tiers: Enhanced at 90 credits / 1,000 characters (≈ $0.19 per minute of speech) and Premium at 180 (≈ $0.37) for the most natural result. The OpenAI Standard tier reads Chinese too at 50 credits, with 6 delivery styles. Every clip downloads as MP3. Browse all 138 voices →
Why a native Chinese voice matters
Every other language on this site can survive a voice that gets the melody slightly wrong. Chinese cannot: mā, má, mǎ and mà are four different words, and a voice that treats pitch as English sentence intonation does not produce an accent, it produces the wrong words. Beyond the four tones there are rules that apply while speaking rather than while writing. Two third tones in a row change the first one. 一 and 不 shift tone depending on what follows. A voice that learned Chinese from romanized text without those rules sounds mechanically correct and subtly wrong.
Then there are the characters that have more than one reading - 行 is xíng in 行走 and háng in 银行, 长 is cháng or zhǎng, 重 is zhòng or chóng - and only context decides. Native voices resolve them from the sentence. All four here are mainland Mandarin speakers. Stacy is a young, sweet narrator and Amy a friendly conversational voice for dialogue and social video. Jason is a deep male voice for documentary and advertising, James a calm one for long-form reading and explainers.
How it works
Paste your text
Up to 20,000 characters - about 20 minutes of speech. Paste Chinese text and the app detects the language and puts the Chinese voices first.
Pick a voice
Every voice has a free sample in its own language - audition before spending a credit.
Play & download
Ready in seconds. Listen in the browser or download the MP3; clips stay in your history for 90 days.
Frequently asked questions
Is this Mandarin or Cantonese?
Mandarin - standard putonghua, the register used in broadcasting on the mainland. Cantonese is a different language and Speecho does not offer it; a Mandarin voice reading Cantonese text would be wrong in a way no setting can fix.
Simplified or traditional characters?
Both are read correctly - the voices work from the characters, not from a particular character set. What they produce is Mandarin pronunciation either way, so traditional text comes out in Mandarin rather than in a Taiwanese or Hong Kong reading.
Does it get tone sandhi and multi-reading characters right?
Yes, and this is the practical reason to use a native voice. Consecutive third tones shift, 一 and 不 change with what follows, and characters like 行, 长 and 重 are resolved from context rather than from a fixed lookup.
How much does Chinese text to speech cost?
Native voices cost 90 credits per 1,000 characters on Enhanced or 180 on Premium. Chinese is denser than European languages, so 1,000 characters is a long passage - closer to four minutes of speech than one. Billing is per character either way, and the first clip is free.
Give your Chinese text a voice
Preview every voice free - pay only for the characters you convert.
Try Chinese text to speech →