News Anchor
MiniMax Speech 2.8 HD · MiniMax
各位观众,晚上好。今天全国大部分地区天气晴朗,北方地区气温明显回升。
Good evening, everyone. Today most of the country is sunny, and temperatures in the north are rising noticeably.
Mandarin · Cantonese · 6 regional dialects
Paste Chinese text — simplified or traditional — and hear it in Mandarin, Cantonese or a regional dialect: news anchors, radio hosts, storytellers and friendly everyday voices.

Updated ·by the Kenerate AI team
On this page
Quick answer
Updated
Kenerate AI's Chinese text to speech reads simplified or traditional Chinese in Mandarin, Cantonese and regional dialects, for videos, lessons and narration. MiniMax has 33 Mandarin and 9 Cantonese voices, Kling adds dialect voices and a Taiwanese accent, and Qwen3 and ElevenLabs also speak Chinese. Scripts run up to 10,000 characters per clip.
Listen first
Each card shows the exact Chinese script with an English translation. The words light up as they're spoken.
MiniMax Speech 2.8 HD · MiniMax
各位观众,晚上好。今天全国大部分地区天气晴朗,北方地区气温明显回升。
Good evening, everyone. Today most of the country is sunny, and temperatures in the north are rising noticeably.
MiniMax Speech 2.8 HD · MiniMax
夜深了,城市慢慢安静下来。接下来这首歌,送给还在加班的你。
It's late, and the city is growing quiet. This next song is for you, still working overtime.
MiniMax Speech 2.8 HD · MiniMax
大家好!今天教大家做一道十分钟就能完成的番茄炒蛋。
Hi everyone! Today I'll teach you a tomato-and-egg stir-fry you can make in ten minutes.
MiniMax Speech 2.8 HD · MiniMax
喺香港嘅老街入面,有一間開咗五十年嘅茶餐廳。
On an old street in Hong Kong there's a cha chaan teng that has been open for fifty years.
Kling V1 TTS · Kuaishou Kling
哎呀妈呀,老铁,今儿这天儿可真冷,赶紧进屋喝口热乎的!
Oh my, buddy, it's freezing today — hurry inside and have something hot!
Qwen3 TTS · Alibaba Qwen
年轻人,别着急。泡一壶好茶,慢慢聊,事情总会有办法的。
Young man, don't rush. Brew a pot of good tea, take your time — there's always a way.
How to
Step 1
Up to 10,000 characters per clip on MiniMax and ElevenLabs Multilingual v2 (5,000 on Eleven v3 and Qwen3).
Step 2
Filter by Chinese or Cantonese for MiniMax's native voices, pick a Kling dialect voice, or use a Qwen3 voice such as Uncle Fu or Dylan.
Step 3
Emotion, speed and pitch on MiniMax; stability and audio tags on ElevenLabs; a style prompt on Qwen3.
Step 4
MP3 (or WAV/FLAC on MiniMax), no watermark, saved to your history.
See it in the app

Open Voice Studio in Text to speech and type the words you want spoken, or start from a script template.

Open the voice card and choose from hundreds of voices by accent, age and tone.

Pick the speech model — some add emotion and audio tags like pauses, whispers or laughs.

Press Generate speech to turn the script into a voiceover.

Finished clips land in your list with a waveform — play, download or reuse them.
Text to speech, emotions & tags, Hindi, voice design, cloning & chat
Chapters
Dialects
Chinese isn't one sound. These are the varieties the Voice Studio can speak today.
Mandarin普通话
你好,很高兴认识你。
MiniMax (33), Qwen3, Kling, ElevenLabs
Cantonese粵語
你好,好高興識到你。
MiniMax (9 voices)
Northeastern东北话
老铁,今儿可真冷!
Kling · Northeast Buddy
Sichuanese四川话
巴适得板!
Kling · Sichuan Girl
Chongqing重庆话
要得嘛!
Kling · Chongqing Lad
Chaoshan潮汕话
食未?
Kling · Chaoshan Uncle
Tianjin天津话
介嘛呀?
Kling · Tianjin Sister
Taiwanese Mandarin台湾国语
你好喔~
Kling · Taiwan Man
Use cases
Reach Chinese speakers without booking a studio.
Publish a Chinese version of every video with one consistent voice.
Model pronunciation for vocabulary, dialogues and listening practice.
Chinese IVR prompts and greetings for your callers.
Narrators for Chinese fiction, kids' tales and long reads.
Chinese spots and demos for local campaigns.
Audio versions of Chinese notices, articles and documents.
Engines
Facts from the Voice Studio's code. The maker's real model name is shown on every clip.

MiniMax
Most natural MiniMax voice; 7 emotions, (laughs)/(sighs) interjections, pronunciation dictionary
Open MiniMax Speech 2.8 HD
Kuaishou Kling
46 character and dialect voices for short lines
Open Kling V1 TTS
Alibaba Qwen
9 characterful voices plus a free-text style prompt
Open Qwen3 TTS
ElevenLabs
Most expressive; audio tags like [whispers], [laughs], [excited]
Open ElevenLabs Eleven v3Controls
The same controls work in every language.
MiniMax voices. 1× is the natural pace; 0.85–0.95× suits narration and learners, 1.1–1.2× suits ads and recaps.
MiniMax voices, in semitones. A few steps down sounds bigger and older; a few up sounds lighter and younger.
MiniMax: neutral, happy, sad, angry, fearful, disgusted, surprised — the same words, a different read.
MiniMax 2.8 reads 13 cues like (laughs), (sighs), (breath); ElevenLabs Eleven v3 reads audio tags like [whispers], [excited], [sarcastic].
ElevenLabs. Lower = more expressive and varied between takes; higher = steadier and more even.
MiniMax lets you pick the format and sample rate (16–44.1 kHz); every clip downloads without a watermark.
Scripts
Chinese scripts to audition voices with.
广告
本周末全场八折,线上线下同步优惠,千万别错过!
20% off everything this weekend, online and in store — don't miss it!
课程
今天我们来学习四个声调:一声平,二声扬,三声拐弯,四声降。
Today we'll learn the four tones: first flat, second rising, third dipping, fourth falling.
冥想
闭上眼睛,慢慢吸气,再慢慢呼气。让身体放松下来。
Close your eyes, breathe in slowly, then breathe out slowly. Let your body relax.
粤语
多謝你嘅支持,我哋下次再見!
Cantonese: Thanks for your support — see you next time!
Tips
Characters with more than one reading are the main thing to watch.
For polyphonic characters (行, 长, 重), rephrase or add a pronunciation entry in MiniMax 2.8's dictionary.
Use full-width punctuation (,。!?) — it gives the voice natural pauses.
Pick a Cantonese voice for Cantonese text; a Mandarin voice will read Cantonese characters in Mandarin.
Dialect voices (Kling) are best for short, characterful lines up to 512 characters.
Write numbers the way they should be read: 二〇二六年 vs 两千零二十六.
Traditional and simplified characters both work; keep one script per clip.
Example uses
Illustrative examples of typical workflows, not customer reviews.

A food vlogger can narrate a street-snack tour in Mandarin with a warm everyday voice and add a Cantonese version for viewers in Hong Kong.

A Mandarin teacher can generate listening clips that model the four tones for adult learners, slowing the voice for practice sentences.

A small shop owner can record a Mandarin and Cantonese phone menu so callers hear their own language before reaching staff.
FAQ
7 questions
Yes. MiniMax has 9 native Cantonese voices, including a narrator, two professional hosts and a professor. Write the text in Cantonese and pick a Cantonese voice.
Link to this answerMandarin (普通话) is the standard spoken language of mainland China and Taiwan; Cantonese (粵語) is spoken in Hong Kong, Macau and Guangdong. Pick the one your audience speaks.
Link to this answerYes — Kling has Northeastern, Sichuan, Chongqing, Chaoshan and Tianjin dialect voices and a Taiwanese-accent voice, for short lines.
Link to this answerYes, both simplified and traditional characters are read correctly.
Link to this answerRephrase with a synonym, or on MiniMax Speech 2.8 add a pronunciation dictionary entry for that word.
Link to this answerA free Kenerate account starts with free credits for any engine and voice, with no watermark.
Link to this answer把简体或繁体中文文本粘贴到 Kenerate 的 Voice Studio,选择普通话、粤语或方言音色,点击生成,然后下载音频即可。MiniMax 有 33 个普通话音色和 9 个粤语音色,Kling 还提供东北话、四川话等方言音色,适合短句。MiniMax 每段最多 10,000 个字符,可下载 MP3、WAV 或 FLAC。免费账户自带免费额度,音频没有水印。
Link to this answerIn depth
What works, what to avoid, and how the pieces fit.
“Chinese” covers several mutually unintelligible varieties — Mandarin, Cantonese (Yue), Wu, Min and more [1]. Standard Mandarin is the official spoken standard [2], while Cantonese is the everyday language of Hong Kong and Guangdong [3]. A text-to-speech voice is built for one of them, so match the voice to the variety.
Mandarin has four tones plus a neutral tone, and Cantonese has six; the same syllable with a different tone is a different word [4]. Tones also change next to each other (tone sandhi), and many characters have more than one reading. Neural voices learn these patterns from context, which is why natural phrasing and correct punctuation matter.
A quick map:
Keep going
By the Kenerate AI team·Last updated and reviewed
We build and run the Kenerate Voice Studio. Every demo on this page is a real output of the engine and voice named on it, made from the script shown; engine facts were checked against the app's code.
Sources: Kenerate Voice Studio (engines, voices and limits from the app's code) · MiniMax Speech 2.8 HD on WaveSpeed · ElevenLabs Eleven v3 on WaveSpeed · ElevenLabs — text to speech documentation · Wikipedia — Varieties of Chinese · Wikipedia — Standard Chinese · Wikipedia — Cantonese · Wikipedia — Tone (linguistics)

Free to start. No watermark. Download every clip.
Tried it? Tell us how it went