Not Mandarin. Not generic “Chinese.” Cantonese — Hong Kong rhythm and Traditional Chinese input, powered by Huihui Neural (zh-HK-HuihuiNeural). Preview, tune speed and pitch, download MP3.
Enter your text above and click to generate natural speech
Most “Chinese” TTS tools are really Mandarin engines: they apply Mandarin phonology to your characters and sound wrong to Cantonese speakers — not “accented Cantonese,” but a different language behind the same script.
This page is tuned for Hong Kong Cantonese (zh-HK) with neural voices such as Huihui: nine tones, Cantonese vowels and consonants, HK pacing, and support for real written Cantonese including colloquial particles. If you have ever pasted Cantonese into a generic Chinese TTS and winced at the result, you already know why that distinction matters.
Mandarin has four lexical tones; Cantonese has nine. Applying Mandarin tone rules to Cantonese text creates wrong words and wrong meanings — not a “Cantonese accent.”
Colloquial written Cantonese uses particles and structures that do not exist in Mandarin. A Mandarin-first stack cannot model them faithfully.
Many Traditional characters map to different pronunciations in Cantonese vs Mandarin. Mandarin mapping on Traditional text mispronounces before tone errors even show up.
Pick the zh-HK neural voice closest to your register after the list loads — male/female variants and formality differ by voice ID.
Clear, neutral HK delivery for narration and subtitles.
Relaxed pacing and everyday rhythm for social and community content.
Measured delivery for corporate, education, and official scripts.
Warm, contemporary female delivery for accessibility and customer-facing audio.
Grounded male delivery for documentary and professional VO.
Slower, more formal read for heritage, news-style, or older audiences.
Traditional characters, mixed registers, or classroom-style sentences — up to 800 characters per generation.
Start from Huihui Neural (zh-HK), then switch to another Cantonese-capable voice if listed.
Preview, adjust speed/pitch, export MP3 for editors, hosts, or LMS.
| Feature | This tool — Cantonese | Generic “Chinese” TTS |
|---|---|---|
| Tonal system | Cantonese tones | Mandarin (4 tones) |
| Phonological rules | Cantonese-specific | Mandarin-specific |
| Character mapping | Traditional → Cantonese | Often Mandarin-first |
| Colloquial Cantonese | Designed for real HK text | Often weak / wrong |
| Best for | Cantonese speakers & learners | Mandarin-first projects |
HK creators — YouTube, Instagram, TikTok narration for local and diaspora audiences.
Diaspora orgs — community audio for members more comfortable in Cantonese than English.
Learners — reliable nine-tone reference listening between lessons.
Product teams — prompts, tutorials, and accessibility for HK/Macau shipping apps.
Accessibility publishers — spoken versions of news and civic information.
Enterprises in HK/Macau — internal training and customer-facing audio that must sound Cantonese-first, not Mandarin-dubbed.
