tts / README.md
D3vShoaib's picture
fixes
70c75bd
---
title: Pocket TTS (CPU)
emoji:
short_description: A 100M parameters text-to-speech (TTS) model by Kyutai-Labs
license: apache-2.0
colorFrom: green
colorTo: green
sdk: gradio
sdk_version: 6.3.0
app_file: app.py
pinned: true
---
# PocketTTS Gradio Interface
This repository contains a **Gradio-based web interface** for [PocketTTS](https://github.com/kyutai-labs/pocket-tts), a lightweight, CPU-optimized Text-to-Speech model by Kyutai Labs.
## 🚀 Key Features
- **Fast & Lightweight**: 100M parameter model optimized for CPU.
- **Low Latency**: ~200ms first chunk on local installs.
- **Multiple Voices**: Choose from voices like `alba`, `marius`, `javert`, and more.
- **Clean UI**: A professional and pretty interface built with Gradio.
## 🛠️ Installation
1. **Clone the repository**:
```bash
git clone https://github.com/D3vShoaib/pocket-tts
cd pocket-tts
```
2. **Install dependencies**:
Make sure you have Python 3.10+ and PyTorch 2.5+ installed.
```bash
pip install gradio pocket-tts torch
```
3. **Run the application**:
```bash
python app.py
```
## 🔊 Available Voices
The app supports the following built-in voices:
- `alba` (Default)
- `marius`
- `javert`
- `jean`
- `fantine`
- `cosette`
- `eponine`
- `azelma`
## 🔗 Links
- **Official Demo**: [kyutai.org/tts](https://kyutai.org/tts)
- **GitHub**: [kyutai-labs/pocket-tts](https://github.com/kyutai-labs/pocket-tts)
- **Model Card**: [kyutai/pocket-tts](https://huggingface.co/kyutai/pocket-tts)
- **Paper**: [arXiv:2509.06926](https://arxiv.org/abs/2509.06926)
## ❤️ Credits
Built by [D3vShoaib](https://github.com/D3vShoaib).
_Note: This is an independent demonstration of the PocketTTS model and is not officially affiliated with Kyutai Labs._