Can I generate Cantonese instead of Mandarin?

Choose SloppyDog 香港粵語 for Cantonese or Joanne Mandarin for Mandarin. A shared Chinese language code does not choose the dialect for you, so listen to the sample and select the matching voice.

Does the tool handle mixed Chinese and English text?

Try a short mixed-language passage first, as pronunciation can vary by voice. The tool does not translate your text; use a separate English voice for an English section if the chosen Chinese voice does not sound suitable.

Is there a free plan for Chinese TTS?

Yes. Sign in with Google and check in daily to claim 1,500 characters. Characters expire at 00:00 UTC, and shared service limits may also restrict generation.

Do I need a credit card to use the free tier?

No. The free tier needs only a Google sign-in and a daily check-in; no credit card is required to generate and download audio.

Do I need to install any software?

No. Voxelle runs entirely in the browser — type text, pick a voice, generate, and download the MP3. There is nothing to install.

Which model and voices does it use?

Speech is generated by the Fish Audio s2.1-pro model. The licensed catalog currently lists 82 voices across 13 languages; Voxelle does not train its own voice models.

What is the per-request character limit?

Each generation request is limited to 1,000 characters. The daily check-in allowance is 1,500 characters, resets at 00:00 UTC, and does not roll over.

Yes. Signed-in users can mint an API key under API keys and call the authenticated REST endpoint to generate speech programmatically.

AI voice studio

chinese text to speech

Enter Chinese text and compare nine 中文 voices. For Mandarin, try Joanne Mandarin; for Cantonese, choose SloppyDog 香港粵語. Listen to each public sample before generating your own script.

Voxelle converts Chinese text to speech with the Fish Audio s2.1-pro model, with 中文 voices you can preview and download as MP3. The free tier needs a Google sign-in and a daily check-in for 1,500 characters (resets at 00:00 UTC, no rollover); each request is capped at 1,000 characters. Commercial use requires a paid plan.

Start creating

01Your wordsTEXT

Every great story
deserves to be heard.

AI voice synthesis

02A voice with feelingAUDIO

Text in. A little more human, out.

01 / Voice studio

Voice studio

Your words stay yours. Text is sent only when you generate audio.

Your script

33 / 300 characters

CJK characters count ×3

Free preview: up to 300 characters, no account needed. Your text is sent to our cloud voice provider to synthesize audio and is not saved. Sign in to save history and unlock longer scripts.

Your composition

Your generated voice will appear here.

Write a short passage and press generate to begin.

Voice

LanguageVoice

Multi-speaker dialogue

ModelSpeed1.00×Format

Inline markers like (whispers) or (excited) steer delivery.

Natural expression

Find the right rhythm for every idea.

Stay in the flow

No complex setup, just a clear next step.

Take it with you

Listen, refine, and download your voice.

How Chinese text to speech works

Paste Chinese text into the editor, choose a voice, and listen to its public sample. Sign in with Google, check in to claim characters, then generate and download an MP3. Samples demonstrate the voice; they are not generated from your script.

Joanne Mandarin
female · middle-aged · narration · educational
Public provider sample
沉稳男声
male · middle-aged · narration · educational
Public provider sample
SloppyDog 香港粵語
male · young · narration · educational
Public provider sample

Writing for Chinese speech

Choose a Mandarin voice or the explicitly named 香港粵語 voice before generating. For Cantonese, try a short sentence written in spoken Cantonese rather than assuming a Mandarin script will be translated.

把文字变成声音，内容就能被更多人听见。输入文本，然后选择一个声音。

Free allowance and audio permissions

Daily check-in grants 1,500 characters per day. Credits expire at 00:00 UTC and do not roll over. Each request is limited to 1,000 characters; shared service budgets and human verification also apply.

Free tier: personal evaluation only. Paid plans include a commercial licence for generated audio. The privacy policy explains text processing and sample playback.
Terms of Service · Privacy Policy

Ways to use Chinese TTS

Video and podcast narration — turn a script into a voice track.
Learning and product walkthroughs — test pronunciation on a short passage before recording a whole course.
Accessibility — listen to documents and articles.
Pronunciation practice — listen to a word or sentence before repeating it.

Frequently asked questions

Can I generate Cantonese instead of Mandarin?: Choose SloppyDog 香港粵語 for Cantonese or Joanne Mandarin for Mandarin. A shared Chinese language code does not choose the dialect for you, so listen to the sample and select the matching voice.
Does the tool handle mixed Chinese and English text?: Try a short mixed-language passage first, as pronunciation can vary by voice. The tool does not translate your text; use a separate English voice for an English section if the chosen Chinese voice does not sound suitable.
Is there a free plan for Chinese TTS?: Yes. Sign in with Google and check in daily to claim 1,500 characters. Characters expire at 00:00 UTC, and shared service limits may also restrict generation.
Do I need a credit card to use the free tier?: No. The free tier needs only a Google sign-in and a daily check-in; no credit card is required to generate and download audio.
Do I need to install any software?: No. Voxelle runs entirely in the browser — type text, pick a voice, generate, and download the MP3. There is nothing to install.
Which model and voices does it use?: Speech is generated by the Fish Audio s2.1-pro model. The licensed catalog currently lists 82 voices across 13 languages; Voxelle does not train its own voice models.
What is the per-request character limit?: Each generation request is limited to 1,000 characters. The daily check-in allowance is 1,500 characters, resets at 00:00 UTC, and does not roll over.
Is there an API?: Yes. Signed-in users can mint an API key under API keys and call the authenticated REST endpoint to generate speech programmatically.