README.md

---
license: apache-2.0
language:
- en
base_model:
- parler-tts/parler-tts-mini-v1
pipeline_tag: text-to-speech
---

# Parler TTS Mini v1 GGUF

This repository contains the GGUF files for Parler TTS Mini v0.1.
You can run this model by using [TTS.cpp](https://github.com/mmwillet/TTS.cpp).
**Note**: TTS.cpp currently only compiles on MacOS with Metal support.

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

## Model Information

Model Checkpoint: [parler-tts/parler-tts-mini-v1](https://huggingface.co/parler-tts/parler-tts-mini-v1)

Voice Prompt: "female voice"

## Sample

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

### FP32

<audio controls src="https://huggingface.co/ecyht2/parler-tts-mini-v1-GGUF/resolve/main/sample/fp32.wav"></audio>

### FP16

<audio controls src="https://huggingface.co/ecyht2/parler-tts-mini-v1-GGUF/resolve/main/sample/fp16.wav"></audio>

### Q4_0

<audio controls src="https://huggingface.co/ecyht2/parler-tts-mini-v1-GGUF/resolve/main/sample/Q4_0.wav"></audio>

### Q5_0

<audio controls src="https://huggingface.co/ecyht2/parler-tts-mini-v1-GGUF/resolve/main/sample/Q5_0.wav"></audio>

### Q8_0

<audio controls src="https://huggingface.co/ecyht2/parler-tts-mini-v1-GGUF/resolve/main/sample/Q8_0.wav"></audio>

### 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
```