# Speak

> Speak is a paid MCP server for design and media. Pay-per-call text-to-speech for AI agents: 38 voices, 7 languages, MP3 output. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Paid
- **Platforms:** API
- **Website:** https://speak.kaneky.dev
- **MCP Registry name:** dev.kaneky/speak
- **Categories:** design and media
- **Listing updated:** 2026-11-15

## About

Speak charges $0.004 for each call of up to 1,000 characters, which works out to four dollars per million characters. Longer text is split at sentence ends and sent as several calls. Payment is made in USDC over the x402 protocol, with no account and no API key.

The 38 voices come from the open Kokoro 82M model, covering American and British English, Spanish, French, Hindi, Italian and Brazilian Portuguese. There is no voice cloning. Neither the text nor the audio is stored: speech is generated and returned in the same response.

On the MCP endpoint, initialize and tools/list are free while tools/call is paid. Refused text, failed generation, or a moment when every voice worker is busy returns an error and is not charged. A plain HTTP POST route is also available, and machine-readable descriptions are published as openapi.json, llms.txt and a .well-known/x402 file.

## Features

- **Per-call pricing:** $0.004 per call of up to 1,000 characters, paid in USDC via x402.
- **38 stock voices:** Voices ship with the open Kokoro 82M model across six languages and accents.
- **No voice cloning:** Nobody's voice can be copied through the service.
- **Nothing stored:** Text and audio are returned in the same response and not kept.
- **Errors are free:** Refused text, failed generation or busy workers are not charged.
- **MCP and HTTP:** One speak tool over streamable HTTP, plus a plain POST route.

## Good for

- Give an agent a voice for reading short replies aloud
- Generate MP3 narration for snippets of text
- Add spoken output to an agent without managing a speech account
- Produce Spanish, French or Hindi audio from text

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http speak https://speak.kaneky.dev/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://speak.kaneky.dev/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "speak": {
      "url": "https://speak.kaneky.dev/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "speak": {
      "type": "http",
      "url": "https://speak.kaneky.dev/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "speak": {
      "serverUrl": "https://speak.kaneky.dev/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "speak": {
      "type": "http",
      "url": "https://speak.kaneky.dev/mcp"
    }
  }
}
```

## Tools

- `speak`: Turns up to 1,000 characters of text into an MP3 using one of 38 voices.

## FAQ

### How much does a call cost?

$0.004 per call of up to 1,000 characters.

### Do I need an account or key?

No. Agents pay per call in USDC over x402.

### Which languages are supported?

American and British English, Spanish, French, Hindi, Italian and Brazilian Portuguese.

### Is my text stored?

No. Neither the text nor the audio is stored.

### What happens with text over 1,000 characters?

It is split at sentence ends and sent as several calls.

### Is Speak free?

No. Speak is a paid product.

### How do I add Speak to Claude Code?

Run this command in your terminal: claude mcp add --transport http speak https://speak.kaneky.dev/mcp

### What are the best alternatives to Speak?

The closest alternatives to Speak are HyperFrames by HeyGen, ffmpeg-skill, AvatarSay — Fictional Character & Talking Video MCP, ChartAnt and VfxSmith.

## Alternatives

- [HyperFrames by HeyGen](https://ai-powered.com/mcp/hyperframes-by-heygen): HyperFrames by HeyGen is a freemium MCP server for design and media.
- [ffmpeg-skill](https://ai-powered.com/mcp/ffmpeg-skill): ffmpeg-skill is an open-source MCP server for design and media.
- [AvatarSay — Fictional Character & Talking Video MCP](https://ai-powered.com/mcp/avatarsay-fictional-character-and-talking-video-mcp): AvatarSay — Fictional Character & Talking Video MCP is a freemium MCP server for design and media.
- [ChartAnt](https://ai-powered.com/mcp/chartant): ChartAnt is a freemium MCP server for design and media.
- [VfxSmith](https://ai-powered.com/mcp/vfxsmith): VfxSmith is a freemium MCP server for design and media.
- [CalderFlow](https://ai-powered.com/mcp/calderflow): CalderFlow is a free MCP server for design and media.
- [Eikino](https://ai-powered.com/mcp/eikino): Eikino is a freemium MCP server for security.
- [Hourlore](https://ai-powered.com/mcp/hourlore): Hourlore is a freemium MCP server for design and media.

---
Source: https://ai-powered.com/mcp/speak-2
