Indri 0.1 124m Tts GGUF

🚀 Indri GGUF Inference

This project enables running Indri models in GGUF format on CPU. Refer to the original model for more details.

🚀 Quick Start

This guide will help you run Indri models on CPU in GGUF format.

📦 Installation

Step 1: Build llama.cpp

To run the inference locally, you need to build the llama.cpp project. The updated guide to do so can be found here.

git clone https://github.com/ggerganov/llama.cpp
cd llama.cpp

cmake -B build
cmake --build build --config Release

Step 2: Download the model

Download the GGUF format models from HuggingFace and place them inside llama.cpp/models/. The models can be found here. Once the model is placed inside the directory, run the llama-cpp server from inside the llama.cpp directory.

# For F16 model, update for different quantization accordingly
./build/bin/llama-server -m /indri-0.1-124M-tts-F16.gguf --samplers 'top_k;temperature' --top_k 15

Refer here if you are facing issues in running the llama-server locally.

Step 3: Run the inference script

Clone the GitHub repository:

git clone https://github.com/cmeraki/indri.git
cd indri

python -m src.tts_gguf --text 'hi my name is Indri' --speaker '[spkr_63]' --out out.wav

Speakers are available here.

You can also run an inference server:

pip install -r requirements.txt

# Install ffmpeg (for Mac/Windows, refer here: https://www.ffmpeg.org/download.html)
sudo apt update -y
sudo apt upgrade -y
sudo apt install ffmpeg -y

python -m server_ggpuf

Redirect to http://localhost:8000/docs to see the API documentation and test the service.

📄 License

This project is licensed under the CC BY-SA 4.0 license.

We are working on making this process more straightforward. Stay tuned for updates!