pnnbao97/VieNeu-TTS

Vietnamese TTS with instant voice cloning • On-device • Real-time CPU inference • 24kHz audio quality • Chuyển văn bản thành giọng nói tiếng Việt • Text to speech tiếng Việt • TTS tiếng Việt

What it solves

VieNeu-TTS provides high-fidelity, natural-sounding text-to-speech (TTS) specifically optimized for the Vietnamese language, including bilingual English-Vietnamese code-switching. It addresses the need for on-device, low-latency speech synthesis that can handle multi-speaker conversations and instant voice cloning without requiring massive computing resources.

How it works

The project offers multiple model architectures (v1, v2, and v3 Turbo). The latest v3 Turbo version uses a from-scratch architecture with a 48 kHz high-fidelity audio output and a dedicated phonemizer. It supports different backends: a torch-free ONNX Runtime for fast CPU inference and streaming, and a PyTorch engine for high-throughput batched generation on GPUs. It also includes a denoising pipeline to clean reference audio clips for voice cloning.

Who it’s for

  • Developers building interactive Vietnamese AI applications or real-time voice assistants.
  • Content creators needing natural Vietnamese voiceovers for podcasts or dialogues.
  • Edge device engineers looking for lightweight, offline TTS solutions that run efficiently on CPU or Apple Silicon.

Highlights

  • Bilingual Support: Seamless transitions between Vietnamese and English.
  • Instant Voice Cloning: Clone a voice using only a 3-8 second audio clip.
  • High Fidelity: v3 Turbo supports 48 kHz audio and experimental emotion cues (e.g., chuckle, sigh).
  • Flexible Deployment: Available as a Python SDK, a Web UI, or a high-performance Docker-based API server.
  • Optimized Performance: Supports frame-level streaming for real-time interaction and batched generation for bulk synthesis.

Related

  • Project
    tronghieuit/v-tts

    A lightweight Vietnamese text-to-speech system featuring multi-speaker synthesis and zero-shot voice cloning that runs efficiently on CPU.

  • Project
    neuphonic/neutts

    A collection of open-source, on-device text-to-speech models that enable natural-sounding speech and instant voice cloning locally on mobile and embedded devices.

  • Project
    Audio8-AI/Audio8_TTS

    A compact 0.6B-parameter multilingual text-to-speech model featuring zero-shot voice cloning and efficient deployment options for CPU and GPU.

  • Project
    Zyphra/ZONOS2

    ZONOS2 is a high-fidelity, multilingual text-to-speech model using a Mixture-of-Experts backbone to deliver naturalistic voice cloning and expressive emotional control.

  • Project
    netease-youdao/Confucius4-TTS

    Confucius4-TTS is an LLM-based text-to-speech engine that enables zero-shot, cross-lingual voice cloning across 14 languages without requiring reference transcripts.