dTelecom x402 Gateway
Pay-per-use communication and AI services for agents via x402 and MPP payment protocols
Built on dTelecom DePIN — a decentralized real-time communication network. No API keys, no accounts. Any AI agent with a crypto wallet can discover, pay, and use these services autonomously. Accepts USDC on Solana/Base, USDC on Tempo (MPP), and USDT on TRON via Bank of AI.
Services
WebRTC audio
SFU room access for audio applications. Each participant is metered separately; no resolution-based video tariff is offered.
POST /v1/webrtc/token
Streaming STT
Parakeet Ultra with Whisper fallback; 99+ language routes. Optional partial transcripts and Smart Turn. PCM16 mono 16 kHz input.
POST /v1/stt/session
Streaming TTS
Kokoro + Supertonic: 32 languages, 54 voices + 10 styles. SSML and optional AudioSeal watermark. PCM16 mono 48 kHz output.
POST /v1/tts/session
File transcription + diarization
Whisper transcription with optional anonymous speaker labels included. Upload up to 256 MiB / 2 hours. Private results for 24 hours.
POST /v1/transcriptions/jobs
Text/SSML to WAV
Private 48 kHz mono PCM16 WAV download. Up to 10,000 source and expanded characters; optional watermark. Charged on completion.
POST /v1/tts/jobs
Translation
TranslateGemma 4B + Hy-MT2 1.8B, 45 language/script entries. Up to 500 code points; pairs must share an enabled model.
POST /v1/translation
Vision snapshots
Upload a static JPEG, PNG or WebP image and optional question. Up to 6 MiB / 4 million pixels; bounded 96-token answer.
POST /v1/vision/snapshot
Audio moderation
Upload raw PCM16 mono 16 kHz chunks up to 30 seconds. HTTP or ticket-authenticated WebSocket; caller controls recording and actions.
POST /v1/moderation/audio
Image/frame moderation
Static JPEG, PNG or WebP up to 6 MiB / 4 million pixels. Caller samples video frames; not a full-video upload API.
POST /v1/moderation/video-frame
Agent sessions combine WebRTC, STT and TTS. Example: approximately $0.0064/minute for two audio participants, one STT session and 367 spoken TTS characters/minute, excluding the LLM. Actual charges follow usage.
Speech controls: partials, Smart Turn, SSML and optional watermark · Complete API contracts and limits
How It Works
Buy Credits
Send an x402 payment to POST /v1/credits/purchase (USDC on Solana/Base), an MPP payment to POST /v1/credits/purchase/mpp (USDC on Tempo), or an x402 payment to POST /v1/credits/purchase/tron (USDT on TRON).
Choose a Service
Create a session or submit an inference request. See each service for reservation, polling and completion rules.
Connect & Use
Use the returned token to connect. Billed for actual usage; unused reservations are released to your prepaid balance. Unused prepaid credits are refundable on request to the paying wallet. Contact: x402@dtelecom.org.
Base USDC live payments have been verified. Solana, Tempo/MPP and TRON integrations are implemented; their remaining live purchase/settlement qualification is pending.
Quick Start
npm install @dtelecom/x402-client viem
import { DtelecomGateway } from "@dtelecom/x402-client";
import { privateKeyToAccount } from "viem/accounts";
const wallet = privateKeyToAccount(process.env.WALLET_PRIVATE_KEY);
const gw = new DtelecomGateway({ account: wallet });
// One call → WebRTC + STT + TTS
const session = await gw.createAgentSession({
roomName: "agent-demo",
participantIdentity: "assistant",
durationMinutes: 5,
clientIdentity: "user-42",
});
// session.webrtc.agent, session.webrtc.client, session.stt, session.tts — ready to connect
Build voice agents with @dtelecom/agents-js — includes DtelecomSTT and DtelecomTTS providers.
Built With This Stack
WebRTC + STT + TTS powered by x402 · Source on GitHub