Vietnamese Text to Speech: Hanoi and Saigon Voices

For Vietnamese text to speech, paste your script with full tone marks, choose a Northern (Hanoi) or Southern (Saigon) voice and generate. Voxmint offers one voice for each accent, so you can match the audio to the people you are speaking to. This guide covers how to choose the accent, prepare your text and get a natural result.

Try Vietnamese text to speech

Vietnamese is a tonal language, and its accents differ more than most people expect. That is why the first decision is about audience rather than technology. Voxmint has two ready-made Vietnamese voices on its Vietnamese text to speech page, where you can hear a preview of each:

Linh: a gentle Northern presenter voice from Hanoi, soft and precise.
Minh: a relaxed, friendly Southern voice from Saigon.

Both read standard Unicode Vietnamese exactly as you type it.

Northern or Southern: which should you choose?

Northern and Southern Vietnamese differ in pronunciation, in how tones are realised, and in everyday vocabulary. A practical rule of thumb:

Choose the Hanoi voice (Linh) for national campaigns, news-style summaries, corporate video, e-learning and formal announcements. Northern Vietnamese is the broadcast standard across the country.
Choose the Saigon voice (Minh) when you are talking directly to people in the South: social videos, vlogs, food content or a friendly brand voice.

Match your vocabulary to the voice, too. Southern words such as "ly" (glass) or "bông" (flower) sound natural from Minh but odd from a Hanoi voice.

If you need a Central accent, or a younger or older speaker, describe the voice in the Studio instead. For example: "a cheerful young woman from Da Nang in her twenties with a bright, clear voice and a lively pace." Generate a few takes and keep the best one.

How to convert Vietnamese text to speech, step by step

Open the Studio and select Vietnamese.
Choose a voice, Linh for Hanoi or Minh for Saigon, after listening to the previews.
Paste your script with all diacritics (more on that below).
Generate a short test of one or two sentences and listen.
Adjust the text if a word or rhythm is off, then generate the full script.
Download the audio and use it in your video editor or app.

Always include tone marks

Vietnamese has six tones, and the diacritics are what tell a voice which one to use. "ma", "má", "mà", "mả", "mã" and "mạ" are six different words. If you type without tone marks, as people sometimes do in chat, neither a voice nor a human can read the text correctly.

So paste your script in proper Unicode Vietnamese, as any modern keyboard produces it. If your text comes from a document or a website, check that the accents survived the copy and paste.

Writing for the ear

A script that looks fine on paper can sound stiff when read aloud. A few habits help:

Write numbers, dates and currency as you want them spoken. Use "mười lăm tháng ba" instead of 15/3, and "năm mươi nghìn đồng" instead of 50.000đ.
Use short sentences. Commas and full stops control the pacing, and short sentences give the most natural rhythm.
Spell out abbreviations the first time, and write foreign brand names the way you want them pronounced.
Read it aloud yourself. Anywhere you stumble, a voice will probably stumble too.

Honest limits: check the tones

On everyday sentences the tones generally come out right, but a synthetic voice can occasionally flatten or confuse a tone, especially on rare words and names. If a word sounds wrong, generate a fresh take or rephrase the sentence slightly. Listen to anything important before you publish it, particularly names, place names and technical terms.

What people use Vietnamese TTS for

Social video and short-form content: quick voice-overs without recording in a quiet room.
News and corporate video: a calm, consistent narrator for updates and announcements.
E-learning: clear diction for courses, onboarding material and training videos.
Accessibility: audio versions of articles and documents.
Apps and phone menus: voice prompts generated on demand, including through the Voxmint developer API with API keys.

Want your own Vietnamese voice?

Besides the ready-made voices, Voxmint can design a voice from a text description and clone a voice from a few seconds of reference audio. Cloning requires the speaker's consent: only use your own voice or one you have permission to use. Our guide to AI voice cloning explains how it works.

Frequently asked questions

Does Vietnamese text to speech work without diacritics?

No. Tone marks carry meaning in Vietnamese, so text without them cannot be read correctly. Use full Unicode diacritics.

How many Vietnamese voices does Voxmint have?

Two ready-made Vietnamese voices at the moment: Linh (Northern, Hanoi) and Minh (Southern, Saigon). Beyond these, you can describe a voice in words and generate it.

Can I make the voice speak faster or slower?

Pacing is mostly controlled by your writing. Shorter sentences and well-placed commas and full stops change the rhythm more than anything else.

Can I mix English words into a Vietnamese script?

You can, but test it. Write foreign names the way you want them pronounced and listen to a short sample before generating the whole script.

Is the audio ready to publish?

Listen to it first. Synthetic voices are good but not perfect, and a quick review catches the occasional odd tone or name.

Ready to hear it? Try Vietnamese voices on Voxmint and pick the accent your audience expects.