README.md

---
title: ZeroGPU TTS Service
emoji: 🚀
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 5.42.0
app_file: app.py
pinned: false
license: mit
suggested_hardware: zero-a10g
---

# 🚀 ZeroGPU Text-to-Speech Service

A high-performance text-to-speech service powered by **Hugging Face ZeroGPU** and **Nvidia H200** dynamic resources.

## 🎯 Features

- 🚀 **ZeroGPU Acceleration**: Dynamic H200 GPU allocation
- ⚡ **Ultra-Fast Generation**: Optimized for latest GPU architecture
- 🎭 **Multiple Voice Presets**: 10 different voice characteristics
- 🔊 **High-Quality Audio**: Professional-grade speech synthesis
- 📦 **Batch Processing**: Multiple texts in parallel
- 🌐 **Dual Protocol Support**: Gradio Web UI + MCP Protocol
- 🔌 **MCP Integration**: Compatible with AI assistants (Claude Code, etc.)
- 💰 **Cost Efficient**: No idle costs with Pro subscription

## 🏗️ Architecture

- **Backend**: Transformers + PyTorch with CUDA optimization
- **Frontend**: Gradio with enhanced UI
- **GPU**: ZeroGPU with H200 dynamic scaling
- **Model**: Bark (suno/bark-small) with mixed precision

## 🚀 Performance

- **Single synthesis**: 0.5-2 seconds (depending on text length)
- **Batch processing**: Parallel execution on H200
- **Memory efficient**: Automatic GPU cleanup
- **Scaling**: Dynamic resource allocation

## 💻 API Usage

### Gradio Client API
```python
from gradio_client import Client

client = Client("YOUR_USERNAME/tts-gpu-service")
result = client.predict(
    "Hello from ZeroGPU!",
    "v2/en_speaker_6",
    api_name="/predict"
)
audio_file, status = result
```

### MCP Protocol API
```python
from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import stdio_client

async def use_tts():
    server_params = StdioServerParameters(
        command="python",
        args=["app.py", "--mcp-only"]
    )
    
    async with stdio_client(server_params) as (read, write):
        async with ClientSession(read, write) as session:
            await session.initialize()
            
            result = await session.call_tool("tts_synthesize", {
                "text": "Hello from MCP!",
                "voice_preset": "v2/en_speaker_6"
            })
```

## 🔌 MCP Tools

- **`tts_synthesize`**: Convert single text to speech
- **`tts_batch_synthesize`**: Convert multiple texts to speech
- **`tts_get_info`**: Get system status and capabilities

## 🚀 Running Modes

### Dual Mode (Default)
```bash
python app.py
```
- Gradio UI: http://localhost:7860
- MCP Server: Available on stdio

### MCP-Only Mode
```bash
python app.py --mcp-only
```
- For integration with AI assistants
- No web interface, only MCP protocol