vieneu 1.1.4

Advanced on-device Vietnamese TTS with instant voice cloning

VieNeu-TTS

VieNeu-TTS is an advanced on-device Vietnamese Text-to-Speech (TTS) model with instant voice cloning.

[!TIP] Voice Cloning: All model variants (including GGUF) support instant voice cloning with just 3-5 seconds of reference audio.

This project features two core architectures trained on the VieNeu-TTS-1000h dataset:

• VieNeu-TTS (0.5B): An enhanced model fine-tuned from the NeuTTS Air architecture for maximum stability.
• VieNeu-TTS-0.3B: A specialized model trained from scratch, delivering 2x faster inference and ultra-low latency.

These represent a significant upgrade from the previous VieNeu-TTS-140h with the following improvements:

• Enhanced pronunciation: More accurate and stable Vietnamese pronunciation
• Code-switching support: Seamless transitions between Vietnamese and English
• Better voice cloning: Higher fidelity and speaker consistency
• Real-time synthesis: 24 kHz waveform generation on CPU or GPU
• Multiple model formats: Support for PyTorch, GGUF Q4/Q8 (CPU optimized), and ONNX codec

VieNeu-TTS delivers production-ready speech synthesis fully offline.

Author: Phạm Nguyễn Ngọc Bảo

---

---

🔬 Model Overview

• Backbone:
  • VieNeu-TTS (0.5B): Qwen-0.5B fine-tuned from NeuTTS Air.
  • VieNeu-TTS-0.3B: Custom 0.3B model trained from scratch, optimized for extreme speed (2x faster).
• Audio codec: NeuCodec (torch implementation; ONNX & quantized variants supported)
• Context window: 2,048 tokens shared by prompt text and speech tokens
• Output watermark: Enabled by default
• Training data: VieNeu-TTS-1000h — 443,641 curated Vietnamese samples (Used for both versions).

Model Variants

Model	Format	Device	Quality	Speed
VieNeu-TTS	PyTorch	GPU/CPU	⭐⭐⭐⭐⭐	Very Fast with lmdeploy
VieNeu-TTS-0.3B	PyTorch	GPU/CPU	⭐⭐⭐⭐	Ultra Fast (2x)
VieNeu-TTS-q8-gguf	GGUF Q8	CPU/GPU	⭐⭐⭐⭐	Fast
VieNeu-TTS-q4-gguf	GGUF Q4	CPU/GPU	⭐⭐⭐	Very Fast
VieNeu-TTS-0.3B-q8-gguf	GGUF Q8	CPU/GPU	⭐⭐⭐⭐	Ultra Fast (1.5x)
VieNeu-TTS-0.3B-q4-gguf	GGUF Q4	CPU/GPU	⭐⭐⭐	Extreme Speed (2x)

Recommendations:

• GPU users: Use VieNeu-TTS (PyTorch) for best quality
• CPU users: Use VieNeu-TTS-0.3B-q4-gguf for fastest inference or VieNeu-TTS-0.3B-q8-gguf for best CPU quality.
• Streaming: Only GGUF models support streaming inference (Requires llama-cpp-python >= 0.3.16)

---

✅ Todo & Status

• Publish safetensor artifacts
• Release GGUF Q4 / Q8 models
• Release datasets (1000h and 140h)
• Enable streaming on GPU
• Provide Dockerized setup
• Release fine-tuning code (LoRA)
• LoRA Adapter integration in Gradio

---

🌟 New Feature: LoRA Adapters

VieNeu-TTS now officially supports LoRA (Low-Rank Adaptation). This allows you to:

• Use custom fine-tuned voices from Hugging Face.
• Achieve much higher quality and similarity than zero-shot voice cloning.
• Switch between different adapters seamlessly in the Gradio UI.

For more details, see docs/LORA_USAGE.md.

---

🛠️ Fine-tuning

You can now train VieNeu-TTS on your own voice dataset!

• Simple Workflow: Follow the step-by-step guide in finetune/README.md.
• Notebook Support: Use finetune/finetune_VieNeu-TTS.ipynb for an interactive experience.

---

🏁 Getting Started

1. Clone the repository

git clone https://github.com/pnnbao97/VieNeu-TTS.git
cd VieNeu-TTS

2. Install eSpeak NG (Required)

Phonemizer requires eSpeak NG to function.

• Windows: Download installer from eSpeak NG Releases (Recommended: .msi).
• macOS: brew install espeak
• Ubuntu/Debian: sudo apt install espeak-ng
• Arch Linux: paru -S aur/espeak-ng

---

3. Environment Setup (Recommended)

A. Install uv (Fast Python package manager):

• Windows: powershell -c "irm https://astral.sh/uv/install.ps1 | iex"
• Linux/macOS: curl -LsSf https://astral.sh/uv/install.sh | sh

B. Install dependencies:

[!TIP] For NVIDIA GPU Users: To use LMDeploy (Turbo mode), ensure you have updated drivers and CUDA Toolkit 12.8+.

# Default setup (Includes GPU support)
uv sync

# CPU-only mode (No GPU dependencies)
uv sync --no-default-groups

---

4. Quick Start (Web UI)

Once environment is ready, start the Web UI with a single command:

uv run gradio_app.py

Access the Web UI at http://127.0.0.1:7860.

---

---

📦 Using as a Python SDK (via pip)

If you want to integrate VieNeu-TTS into your own project:

1. Windows (Hassle-free setup)

We provide pre-built CPU wheels for llama-cpp-python (version 0.3.16) for Python 3.10 to 3.14 to avoid compilation errors.

pip install vieneu --extra-index-url https://pnnbao97.github.io/llama-cpp-python-v0.3.16/cpu/

2. Linux / macOS / Others

pip install vieneu

3. GPU Support (Remote Server)

For high-performance GPU inference without local complexity, you can set up a remote server using lmdeploy.

A. On the Server (with GPU):

1. Install LMDeploy: pip install lmdeploy[all]
2. Launch the API Server:

lmdeploy serve api_server pnnbao-ump/VieNeu-TTS-0.3B --server-port 23333 --tp 1

B. On the Client (CPU/Laptop): Connect to the server using the SDK:

from vieneu import Vieneu

# Connect to the remote server
tts = Vieneu(mode="remote", api_base="http://your-server-ip:23333/v1")

4. Advanced Usage Example (Full Features)

Here is a comprehensive example showing how to initialize, manage voices, clone custom voices, and control generation.

"""
Demo VieNeuSDK v1.1.3 - Full Features Guide
"""

import time
import soundfile as sf
from vieneu import Vieneu
from pathlib import Path

def main():
    print("🚀 Initializing VieNeu SDK (v1.1.3)...")
    
    # Initialize SDK
    # Default: "pnnbao-ump/VieNeu-TTS-0.3B-q4-gguf" (Speed & CPU Optimized)
    #
    # You can change 'backbone_repo' to balance Quality vs Speed:
    # 1. Better Quality (slower than q4): "pnnbao-ump/VieNeu-TTS-0.3B-q8-gguf"
    # 2. PyTorch 0.3B (Fast, uncompressed): "pnnbao-ump/VieNeu-TTS-0.3B"
    # 3. PyTorch 0.5B (Best Quality, heavy): "pnnbao-ump/VieNeu-TTS"
    # You can also use a GGUF version merged with your own LoRA adapter.
    # See finetuning guide: https://github.com/pnnbao97/VieNeu-TTS/tree/main/finetune
    
    # Mode selection:
    # - mode="standard" (Default): Runs locally using GGUF (CPU) or PyTorch
    # - mode="remote": Connects to the LMDeploy server setup: lmdeploy serve api_server pnnbao-ump/VieNeu-TTS-0.3B --server-port 23333 --tp 1
    
    tts = Vieneu()
    # Or to use Remote mode:
    # tts = Vieneu(mode="remote", api_base="http://localhost:23333/v1")
    # Example for using Q8 for better quality:
    # tts = Vieneu(backbone_repo="pnnbao-ump/VieNeu-TTS-0.3B-q8-gguf")

    # ---------------------------------------------------------
    # PART 1: PRESET VOICES
    # ---------------------------------------------------------
    print("\n--- 1. Available Preset Voices ---")
    available_voices = tts.list_preset_voices()
    print("📋 Voices:", available_voices)
    
    # Select a preset voice
    current_voice = tts.get_preset_voice("Binh")
    print("✅ Selected voice: Binh")


    # ---------------------------------------------------------
    # PART 2: CREATE & SAVE CUSTOM VOICE
    # ---------------------------------------------------------
    print("\n--- 2. Create Custom Voice ---")
    
    # Replace with your actual .wav file path and its exact transcript (including punctuation)
    sample_audio = Path(__file__).parent / "example.wav"
    sample_text = "ví dụ 2. tính trung bình của dãy số."

    if sample_audio.exists():
        voice_name = "MyCustomVoice"
        
        print(f"🎙️ Cloning voice from: {sample_audio.name}")
        
        # 'clone_voice' now supports saving directly with 'name' argument
        custom_voice = tts.clone_voice(
            audio_path=sample_audio,
            text=sample_text,
            name=voice_name  # <-- Automatically saves voice to system
        )
        
        print(f"✅ Voice created and saved as: '{voice_name}'")
        
        # Verify functionality
        print("📋 Voice list after adding:", tts.list_preset_voices())
        
        # Switch to new voice
        current_voice = custom_voice
    else:
        print("⚠️ Sample audio not found. Skipping...")


    # ---------------------------------------------------------
    # PART 3: SYNTHESIS WITH ADVANCED PARAMETERS
    # ---------------------------------------------------------
    print("\n--- 3. Speech Synthesis ---")
    
    text_input = "Xin chào, tôi là VieNeu-TTS. Tôi có thể giúp bạn đọc sách, làm chatbot thời gian thực, hoặc thậm chí clone giọng nói của bạn."
    
    # Generate with specific temperature
    print("🎧 Generating...")
    audio = tts.infer(
        text=text_input,
        voice=current_voice,
        temperature=1.0,  # Adjustable: Lower (0.1) -> Stable, Higher (1.0+) -> Expressive
        top_k=50
    )
    sf.write("output.wav", audio, 24000)
    print("💾 Saved: output.wav")

    # ---------------------------------------------------------
    # CLEANUP
    # ---------------------------------------------------------
    tts.close()
    print("\n✅ Done!")

if __name__ == "__main__":
    main()

Method 2: Automatic with Makefile (Alternative)

Best if you have make installed (standard on Linux/macOS, or via Git Bash on Windows). It handles configuration swaps automatically.

• Setup: make setup
• Run Demo: make demo

Then access the Web UI at http://127.0.0.1:7860.

---

---

🐋 Docker Deployment

For a quick start or production deployment without manually installing dependencies, use Docker.

Quick Start

Copy .env.example to .env

cp .env.example .env

Build and start container

# Run with CPU
docker compose --profile cpu up

# Run with GPU (requires NVIDIA Container Toolkit)
docker compose --profile gpu up

Access the Web UI at http://localhost:7860.

For detailed deployment instructions, including production setup, see docs/Deploy.md.

---

📦 Project Structure

VieNeu-TTS/
├── vieneu_tts/            # Core engine implementation (VieNeuTTS & FastVieNeuTTS)
├── finetune/              # LoRA training pipeline
│   ├── configs/           # Training & LoRA configurations
│   ├── data_scripts/      # Data filtering & VQ encoding tools
│   ├── dataset/           # Training data storage
│   ├── output/            # Saved checkpoints & LoRA adapters
│   └── train.py           # Main training script
├── utils/                 # Text normalization and phonemization logic
├── sample/                # Built-in reference voices (audio + transcript + codes)
├── docs/                  # Detailed documentation for LoRA, Deployment, and Docker
├── examples/              # Usage examples and testing audio references
├── gradio_app.py          # Modern Web UI with LoRA & Streaming support
├── config.yaml            # Model, Codec, and Voice registry
├── pyproject.toml         # Unified dependency management (UV/PIP)
├── Makefile               # Shortcuts for setup and execution
└── docker-compose.yml     # Docker orchestration for CPU/GPU modes

---

📚 References

• GitHub Repository
• Hugging Face Model (0.5B)
• Hugging Face Model (0.3B)
• LoRA Usage Guide
• Fine-tuning Guide
• VieNeu-TTS-1000h dataset

---

📄 License

• VieNeu-TTS (0.5B): Original terms (Apache 2.0).
• VieNeu-TTS-0.3B: Released under CC BY-NC 4.0 (Non-Commercial).
  • This version is currently experimental.
  • Commercial use is prohibited without authorization. Please contact the author for commercial licensing.

---

📑 Citation

@misc{vieneutts2026,
  title        = {VieNeu-TTS: Vietnamese Text-to-Speech with Instant Voice Cloning},
  author       = {Pham Nguyen Ngoc Bao},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/pnnbao-ump/VieNeu-TTS}}
}

🤝 Contributing

Contributions are welcome!

1. Fork the repository
2. Create a feature branch: git checkout -b feature/amazing-feature
3. Commit your changes: git commit -m "Add amazing feature"
4. Push the branch: git push origin feature/amazing-feature
5. Open a pull request

---

📞 Support

• GitHub Issues: github.com/pnnbao97/VieNeu-TTS/issues
• Hugging Face: huggingface.co/pnnbao-ump
• Discord: Join with us
• Facebook: Phạm Nguyễn Ngọc Bảo

---

🙏 Acknowledgements

This project builds upon NeuTTS Air for the original 0.5B model. The 0.3B version is a custom architecture trained from scratch using the VieNeu-TTS-1000h dataset.

---

Made with ❤️ for the Vietnamese TTS community