Parler Tts Mini V1 GGUF

🚀 Parler TTS Mini v1 GGUF

This repository offers the GGUF files for Parler TTS Mini v0.1. It enables text - to - speech conversion, and you can run this model via TTS.cpp.

🚀 Quick Start

Note: TTS.cpp currently only compiles on MacOS with Metal support.

You can use the following command to run the model:

./cli --model-path /model/path/to/gguf_file.gguf --prompt "I am saying some words" --save-path /tmp/test.wav

✨ Features

• Model Compatibility: Works with TTS.cpp.
• Multiple Audio Samples: Offers samples in different formats like FP32, FP16, Q4_0, Q5_0, and Q8_0.

📦 Installation

The installation process mainly involves using TTS.cpp. Follow the instructions in the TTS.cpp repository to set it up and then use the provided command to run the model.

💻 Usage Examples

Basic Usage

The basic usage is shown in the command below, which generates speech from the given text prompt and saves it as a WAV file:

./cli --model-path /model/path/to/gguf_file.gguf --prompt "I am saying some words" --save-path /tmp/test.wav

📚 Documentation

Model Information

• Model Checkpoint: parler-tts/parler-tts-mini-v1
• Voice Prompt: "female voice"

Sample

Here are sample audios generated using the prompt: "This is speech generated by the TTS CPP library."

FP32

FP16

Q4_0

Q5_0

Q8_0

Generation Times

The audio is generated using Intel i5 - 14400 (16) @ 4.700GHz on CPU. The generation time of the audio is shown below.

Testing fp32
real    0m39.792s
user    5m52.071s
sys     0m7.977s

Testing fp16
real    0m53.419s
user    8m13.402s
sys     0m7.623s

Testing Q4_0
real    5m24.418s
user    50m57.182s
sys     0m27.437s

Testing Q5_0
real    0m44.292s
user    6m40.970s
sys     0m7.971s

Testing Q8_0
real    0m40.479s
user    5m58.898s
sys     0m9.100s

📄 License

This project is licensed under the Apache - 2.0 license.