Skip to content

talker93/oneMinTTS

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

32 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Welcome to the oneMinTTS GitHub repository! This project is engineered for high-performance speech synthesis and streamlined model management, now featuring support for HuggingFace 🤗 MMS-TTS models!

🔑 Key Features

  • 小/Light-Weight: A sleek executable under 200kB simplifies setup with all libraries included, perfect for easy deployment and rapid distribution of ONNX voice synthesis applications.

  • 快/Agile-Dev: Designed for agility with one C++ file for model inference and one Python file for model export, facilitating quick prototyping from text input to waveform output.

  • 灵/Flex-Compatibility: Fully compatible across Windows, macOS, and Linux, ideal for both desktop and server environments.

🚀 Getting Started

To begin using oneMinTTS, simply clone this repository to your local machine. Given the all-inclusive nature of the project, you can dive straight into importing your ONNX models or experimenting with pre-trained VITS models without the hassle of additional setups.

  1. Clone the Repository
git clone https://github.com/talker93/oneMinTTS.git
  1. Prepare the Environment
cd oneMinTTS/mms_tts && mkdir build && cd build
cmake .. && make
  1. Run the Synthesizer
cd fewByte
./fewByteTTS

💡 Quick Example

❯ ./fewByteTTS
Enter your model name: e.g. facebook/mms-tts-kor
facebook/mms-tts-ara

You entered: facebook/mms-tts-ara
Enter your text: في صمت الليل، أراقب النجوم، وأسمع همسات الريح تحكي عنك. قلبي ينبض بأسماء لا تُنسى، وذكرياتك تطفو كلمع البرق في سمائي. ظلك الممتد عبر الأزمان، يعانق روحي في لحظة سكون. أبحث عنك في كل زاوية، في كل شذا يعبق من الحدائق المعطرة. سألتقي بك عند منعطف القدر، حيث الأحلام تلتقي بالواقع. في انتظار الفجر الجديد، حيث يتجدد عهدنا مع كل شروق.

You entered: في صمت الليل، أراقب النجوم، وأسمع همسات الريح تحكي عنك. قلبي ينبض بأسماء لا تُنسى، وذكرياتك تطفو كلمع البرق في سمائي. ظلك الممتد عبر الأزمان، يعانق روحي في لحظة سكون. أبحث عنك في كل زاوية، في كل شذا يعبق من الحدائق المعطرة. سألتقي بك عند منعطف القدر، حيث الأحلام تلتقي بالواقع. في انتظار الفجر الجديد، حيث يتجدد عهدنا مع كل شروق.
Using model at: /Users/shanjiang/Desktop/oneMinTTS/mms_tts/build/fewByte/assets/models/facebook_mms-tts-ara.onnx
time spent: 6057903 microseconds

Jobs done.
Output written to facebook_mms-tts-ara.wav

🤗 Huggingface Models Export

This project includes three basic models due to GitHub's file size limits. For additional models:

  1. Install Huggingface Dependencies
pip install transformers==4.39.1 torch==2.2.1 onnx==1.14.1
  1. Select Models

Edit model_name in mms_tts/files/model_download.py with desired models from this list.

model_names = ["facebook/mms-tts-eng", "facebook/mms-tts-kor", "facebook/mms-tts-ara", "facebook/mms-tts-deu", "facebook/mms-tts-hin", 
                "facebook/mms-tts-spa", "facebook/mms-tts-mal", "facebook/mms-tts-kan", "facebook/mms-tts-fra", "facebook/mms-tts-rus"]
  1. Export Models
python files/model_download.py

VITS2

For using the vanilla VITS2 model, refer to the vits2 folder instructions.

🤝 Contributing

Contributions are welcome! Feel free to post questions and suggestions in Issues.

📄 License

Licensed under the MIT License - see the LICENSE file for details.

About

Launch your speech synthesis within one minute.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors