@dtelecom/stt-mcp
v0.2.0
Published
MCP server for dTelecom real-time speech-to-text with x402 micropayments
Readme
@dtelecom/stt-mcp
MCP (Model Context Protocol) server for dTelecom real-time speech-to-text with x402 micropayments.
Lets AI assistants (Claude, Cursor, etc.) transcribe audio files using dTelecom STT — pay-per-use with USDC, no API keys needed.
Tools
| Tool | Description |
|------|-------------|
| transcribe_file | Transcribe a WAV file (PCM16, 16kHz, mono) to text |
| stt_pricing | Get current pricing ($0.005/min) |
| stt_health | Check service health |
Setup
1. Install
npm install -g @dtelecom/stt-mcp2. Get a wallet
You need a private key with USDC. Either:
- EVM (Base): Ethereum private key (0x hex) with USDC on Base — MetaMask, etc.
- Solana: Solana private key (base58) with USDC on Solana — Phantom, Solflare, etc.
3. Configure your AI assistant
Claude Code (~/.claude.json):
{
"mcpServers": {
"dtelecom-stt": {
"command": "dtelecom-stt-mcp",
"env": {
"DTELECOM_PRIVATE_KEY": "YOUR_PRIVATE_KEY"
}
}
}
}Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"dtelecom-stt": {
"command": "npx",
"args": ["-y", "@dtelecom/stt-mcp"],
"env": {
"DTELECOM_PRIVATE_KEY": "YOUR_PRIVATE_KEY"
}
}
}
}Cursor (Settings > MCP Servers > Add):
{
"dtelecom-stt": {
"command": "npx",
"args": ["-y", "@dtelecom/stt-mcp"],
"env": {
"DTELECOM_PRIVATE_KEY": "YOUR_PRIVATE_KEY"
}
}
}4. Convert audio (if needed)
The tool accepts WAV files in PCM16 16kHz mono format. Convert with:
ffmpeg -i input.mp3 -ar 16000 -ac 1 -acodec pcm_s16le output.wavEnvironment Variables
| Variable | Required | Default | Description |
|----------|----------|---------|-------------|
| DTELECOM_PRIVATE_KEY | Yes | — | EVM key (0x hex) or Solana key (base58) |
| DTELECOM_STT_URL | No | https://x402stt.dtelecom.org | STT service URL |
Pricing
- $0.005/min, billed per session
- Minimum 5 minutes ($0.025)
- Paid in USDC on Base or Solana via x402 protocol
- No accounts, no API keys, no subscriptions
Links
License
MIT
