Vietnamese TTS: Getting Tones Right and Choosing an Accent
To get Vietnamese tones right in text to speech, paste your script with every tone mark in place and check the spellings that are easy to mix up, such as hỏi and ngã. Then choose the accent by audience: the Northern (Hanoi) voice for national and formal content, the Southern (Saigon) voice for people in the South. You can hear both on the Vietnamese text to speech page. This guide covers the tone and spelling problems that cause most bad reads, and how to pick between the two accents.
Why tone marks decide the word
Vietnamese has six tones, and five of them are written with a mark. The voice reads the mark, so the mark decides the word:
| Tone | Mark | Example | Meaning |
|---|---|---|---|
| ngang | none | ma | ghost |
| huyền | grave | mà | but |
| sắc | acute | má | cheek, or mother in the South |
| hỏi | hook above | mả | grave, tomb |
| ngã | tilde | mã | horse, or code |
| nặng | dot below | mạ | rice seedling |
Without marks, a single syllable can stand for several words. "ban" could be bàn (table), bán (sell), bạn (friend) or bản (a copy or version). A sentence typed the way people sometimes type in chat, such as "toi muon mua ban", cannot be read correctly by a person or a voice.
Check your script before you generate
Many tone problems start in the text. A few checks catch them.
Marks lost in copy and paste
If your script comes from a PDF, a website, a spreadsheet or an old document, check that the diacritics survived. Look for words that suddenly appear without marks, or with a mark on the wrong letter. Paste your text in proper Unicode Vietnamese, as any modern keyboard produces it.
Hỏi and ngã
These two tones are the most common spelling mistake in Vietnamese, partly because most Southern speakers pronounce them the same way. In writing they are still different words:
- nghỉ (to rest) and nghĩ (to think)
- sửa (to fix) and sữa (milk)
A voice reads what is written. If the script says "sữa xe" when you meant "sửa xe" (fix the bike), you will hear milk. When a word sounds wrong, check the hỏi or ngã mark first.
Other spellings that sound alike
Some consonants spelled differently sound the same in one accent. Northern speakers, for example, usually pronounce "s" and "x" alike, and "tr" and "ch" alike, so these are frequent spelling slips. Proofread them even if they sound identical to you.
Tone-mark placement
You will see both "hoà" and "hòa", or "thuý" and "thúy", in real Vietnamese text. Both styles are in common use. If a word with a mark on the second vowel reads oddly, try the other placement, generate a short test and listen.
Names, rare words and loanwords
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. Write foreign brand names the way you want them pronounced, and listen to any names, place names and technical terms before you publish.
Northern or Southern: choosing the accent
Northern and Southern Vietnamese differ in pronunciation, in how tones are realised and in everyday vocabulary. Voxmint has one ready-made voice for each:
- Linh, a gentle Northern presenter voice from Hanoi. Choose her for national campaigns, news-style summaries, corporate video, e-learning and formal announcements. Northern Vietnamese is the broadcast standard across the country.
- Minh, a relaxed, friendly Southern voice from Saigon. Choose him when you are talking directly to people in the South: social video, vlogs, food content or a warm brand voice.
Match the vocabulary to the voice
Choosing an accent also means choosing words. Southern vocabulary sounds natural from Minh but odd from a Hanoi voice, and the reverse is true too:
| Meaning | Northern (Linh) | Southern (Minh) |
|---|---|---|
| glass | cốc | ly |
| flower | hoa | bông |
| fruit | quả | trái |
| pig, pork | lợn | heo |
| spoon | thìa | muỗng |
| dad | bố | ba |
| thousand | nghìn | ngàn |
The friendly sentence-final particles change as well. A Northern script might end "nhé", where a Southern one would use "nha" or "nè". Minh's own preview line shows the Southern style: "Chào mấy bạn nha! Sáng nay Sài Gòn nắng đẹp quá trời…"
Numbers and prices
Write numbers, dates and currency as you want them spoken, and in the vocabulary of the accent. For a Hanoi read, "năm mươi nghìn đồng" works better than 50.000đ. For Minh, "năm mươi ngàn đồng" sounds more at home. Write "mười lăm tháng ba" instead of 15/3.
If neither accent fits
There is no ready-made Central voice. You can describe one 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." Results vary, so generate a few takes and compare them. For a long series where the speaker must never change, Linh and Minh are the safer choice, because catalogue voices are fixed to the same speaker.
A quick test before the full script
- Proofread the tone marks, especially hỏi and ngã.
- Generate one or two sentences that contain your trickiest names and terms.
- Listen for any tone that sounds flat or wrong, and fix the spelling or rephrase.
- Generate the full script, then listen once more before you publish.
Frequently asked questions
Does Vietnamese text to speech work without tone marks?
No. Tone marks carry meaning, so text without them cannot be read correctly. Use full Unicode diacritics.
A word sounds like the wrong tone. What should I check first?
Check the spelling, especially hỏi and ngã, and whether the marks survived copy and paste. If the spelling is correct, generate a fresh take or rephrase the sentence slightly.
Can I use Southern words with the Hanoi voice?
You can, but it will sound a little unusual, as it would from a real Hanoi speaker. Use Southern words with Minh and Northern words with Linh.
Which accent should I use for a national audience?
Northern. Linh's Hanoi accent is the broadcast standard, so it suits national campaigns, e-learning and formal communication.
Hear both accents on the Vietnamese text to speech page and pick the one your audience expects.