# Strands Agents text to speech through MCP

A Strands agent that answers in more than one language needs a voice for each of them. Load the Paxa MCP server through the MCP client built into Strands, and the agent speaks English, Thai, and Mandarin in any Paxa TTS Flash v2 voice. The same server gives it transcription, OCR, and translation tools.

## The Paxa MCP server as Strands tools

The agent starts the server with npx on first use and stops it when it is done. A server-side agent keeps only the file tools with tool_filters.

## From a key to an agent with a voice

Strands needs no Paxa-specific package. Its MCP client loads the tools, and the agent keeps its own model provider.

1. **Install Strands**: Install strands-agents for Python or @strands-agents/sdk for TypeScript, install Node.js for npx, and set PAXA_API_KEY to a key from the dashboard.
2. **Load the Paxa tools**: Create an MCP client that runs npx -y @paxalabs/mcp with PAXA_API_KEY in its environment, and pass it to the agent's tools.
3. **Keep the tools you need**: The playback tools, design_voice, refine_voice, and listen use the speakers and microphone of the machine running the server. In Python, pass tool_filters to keep the file tools.

Paxa TTS Flash v2 lists at $10 per 1M characters. Paxa STT Lite lists at $0.50 per hour of audio, billed by the millisecond.

## Questions

### What is Strands Agents?

Strands describes its SDK as a way for developers to quickly build, manage, evaluate, and deploy AI-powered agents. It is available in Python and TypeScript, and Amazon Bedrock is its default model provider.

### Which Paxa tools does the agent get?

The agent gets text to speech, transcription, OCR, field extraction, and translation tools. [The MCP server page](https://paxalabs.com/docs/mcp#tools) lists every tool and its price, and [the Paxa MCP server](https://paxalabs.com/mcp) works in other MCP clients too.

### Which voices can the agent use?

Speech runs on Paxa TTS Flash v2, whose 50 voices each speak English, Thai, and Mandarin. [Direction tags](https://paxalabs.com/docs/directions) written into the text, such as `[whispering]` or `[laughs]`, set the delivery.

### Where does the agent save speech?

`text_to_speech` saves files to `PAXA_OUTPUT_DIR`, which defaults to the working directory. Set it and `PAXA_DEFAULT_VOICE` in `env` next to `PAXA_API_KEY` ([environment variables](https://paxalabs.com/docs/mcp#environment)).

## Start with free credits

Sign in with Google, GitHub, or Hugging Face and spend 100 one-time free credits on your own text.