Vietnamese at the core
Diacritics, sentence breaks and pacing are tuned for Vietnamese content.
VieNeu v3 Turbo · natural Vietnamese voices
Create natural narration, build long-form content in Studio and integrate the OpenAI-compatible API.
Promotion through 31 Dec 2026 Calculating time remaining…
Listen on this device
Choose a style to hear a VieNeu Turbo sample.
Core value
Diacritics, sentence breaks and pacing are tuned for Vietnamese content.
Split sections, choose voices and export one complete file.
A familiar endpoint with enough documentation to get started in minutes.
Choose the right tool
| Criterion | NitiAI | ElevenLabs | Azure Speech |
|---|---|---|---|
| Focus | Vietnamese | Multilingual | Multilingual |
| Long-form Studio | Yes | Yes | Yes |
| Reading controls | Voice, speed, emotion | Speed, style | Pitch, rate, break |
| Reference price / 1 million characters | Loading… | Public pricing | Public pricing |
| Free plan | Loading… | Current terms | Current terms |
| Source | NitiAI pricing | Official documentation | Microsoft Learn |
Get started
Voice library
Filter by region, gender, age and intended use.
Pricing
Pricing and limits load directly from the service database.
| Plan | Price | Characters | Limits |
|---|
Paid characters are added to your account under the active plan and do not expire each month.
No. Promotional allowances follow the current system configuration.
Yes. AI-generated files include identifying information under NitiAI's AI labeling policy.
API documentation
An OpenAI-compatible API for speech generation and a WebSocket for live streaming.
Replace environment variables with your own values; never put real keys in source code.
401 invalid key · 429 quota exceeded · 422 invalid data.
Text is limited to 5,000 characters per request; Studio accepts files up to 20,000 characters.
Connect wss://tts.niti.vn/v1/audio/stream using the API key in the authorization header.
System status
This status reflects the public console and does not add load to a voice cell.
Please wait.
Legal
By using tts.niti.vn, you confirm that you have rights to the text, images and voice samples you upload; do not use the service for impersonation, fraud or infringement of another person's rights.
Account holders are responsible for protecting API keys. NitiAI may suspend content that violates the law or safety policy.
Users must disclose AI-generated audio where required and must not conceal its origin to cause confusion.
Updated: 27 Oct 2026.
Legal
We collect account information, minimal usage logs and files you actively upload in order to provide the service.
Studio data is retained for up to 30 days through the history feature; you can delete it permanently. API keys are shown once and stored as hashes.
We do not sell personal data. Access, correction or deletion requests: tuyentinh22@gmail.com.
Legal
Create a voice only when you own it or have explicit authorization. Every request must include a consent phrase read by the voice owner and a valid sample clip.
Consent and voice clips are reviewed before use. Voice owners may request deletion; source data is deleted on request and the action is logged.
Impersonation, fraud, harassment, fabricated statements or use of a minor's voice without guardian consent.
Legal
TTS NitiAI labels AI-generated files and returns identifying headers. Users are responsible for clearly disclosing AI content in published products.
These rules follow Law 134/2025 and applicable guidance in Vietnam.
Contact
Technical support, enterprise and legal requests: tuyentinh22@gmail.com.
Service address: https://tts.niti.vn/.
Sign-up requires an OTP sent by email. Google appears only when configured.
Sessions use HttpOnly, Secure, SameSite=Lax cookies. API keys are shown only once.
Studio
Split sentences, choose a voice for each segment, review and export complete audio.
No segments yet.
Export
MP3, WAV, M4A, FLAC or Opus, with SRT/VTT timed from each segment's actual WAV duration.
My voices
You must own the voice or have authorization. Sample clips must be 3–8 seconds.
No consent phrase created.
Nothing recorded.
Studio · custom pronunciation
Set proper names, abbreviations and terminology to sound exactly as you intend. Entries are applied before normalization.
Sign in to open your account's private dictionary.
Sign inW3C PLS pairs grapheme → alias. Import replaces the entire current dictionary in one transaction.