Programmable voice with documentation engineers respect
REST and WebSocket APIs for call control, IVR, recording, conferencing and real-time media streaming — with a free sandbox, idempotent endpoints, signed webhooks and SDKs that do not lag the platform.
// Node.js SDK — place a call and stream media import { Telecorenetwork } from "@telecorenetwork/voice"; const zt = new Telecorenetwork({ apiKey: process.env.ZT_API_KEY }); const call = await zt.calls.create({ from: "+18005550100", to: "+442071234567", answerUrl: "https://app.example.com/ivr", record: true, mediaStream: { url: "wss://app.example.com/media", track: "both" } }); // → { id: "call_7d1f", status: "ringing", latency_ms: 58 } // Signed webhook on answer { "event": "call.answered", "call_id": "call_7d1f", "answered_at": "2026-08-11T09:41:22Z" }
API surface, in full
Every capability in the portal has an API equivalent. There is no feature we hold back for the UI.
Full call control
Create, answer, transfer, bridge, hold, mute, hang up and record — synchronously over REST or reactively through signed webhooks with automatic retry and replay.
IVR and speech
Play audio, collect DTMF, run speech recognition in 30+ languages and synthesise natural text-to-speech, all controllable mid-call without dropping the session.
Real-time media streaming
Bidirectional audio over WebSocket for live transcription, voice-AI agents, sentiment analysis or custom media processing, with sub-100ms added latency.
Conferencing
Programmatic conference rooms with per-participant mute, hold, coach and whisper, plus dial-out, waiting rooms and recording per room or per leg.
SIP REFER and interoperation
Native SIP REFER handling for clean handoff to on-premise systems, so the API integrates with your existing estate instead of replacing it.
Production-grade primitives
Idempotency keys, cursor pagination, RFC-standard rate-limit headers, signed webhooks with timestamp verification and a documented deprecation policy.
Built by people who have integrated other people's telephony APIs
Most telephony APIs are pleasant in the quickstart and painful in production — undocumented rate limits, webhooks that silently stop retrying, breaking changes shipped without notice. We designed against those specific failures.
- Free sandbox with unlimited test calls to reserved numbers — no card, no trial clock
- Idempotency on every mutating endpoint, so retries never double-dial a customer
- Signed webhooks with automatic exponential retry and a replay endpoint for gaps
- Versioned API with a minimum 12-month deprecation window and written migration guides
- Official SDKs for Node, Python, Go, Java, PHP, Ruby and .NET, released with the platform
60ms median latency
Measured API response time at the edge, published on the status page.
7 official SDKs
Versioned together with the API, not maintained by a lone volunteer.
Webhook replay
Recover missed events from an outage window without data loss.
Sandbox included
Unlimited test calls at zero cost, indefinitely.
From API key to production
Most teams have a working proof of concept the same afternoon.
Create a sandbox account
Self-service signup gives you API credentials, reserved test numbers and unlimited test calls in minutes. No sales conversation required.
Build against the reference
Interactive API reference with runnable examples in every SDK language, plus a webhook inspector that shows exactly what we send and what your endpoint returned.
Load test honestly
Request a temporary rate-limit increase on sandbox and test your real concurrency profile before launch, not after your first incident.
Go live with support
Move to production keys, buy numbers, and get a technical contact on the NOC for your launch window at no additional cost.
Core endpoints and events
A representative subset. The full reference documents every endpoint, parameter and error code with runnable examples.
| Endpoint or event | Method | Purpose |
|---|---|---|
| /v2/calls | POST | Place an outbound call with answer URL and recording options |
| /v2/calls/{id} | GET / PATCH | Retrieve call state or modify a live call in progress |
| /v2/calls/{id}/transfer | POST | Blind or attended transfer, including SIP REFER |
| /v2/calls/{id}/stream | POST | Start or stop bidirectional WebSocket media streaming |
| /v2/conferences | POST | Create a conference room with participant policy |
| /v2/recordings | GET | List, download or delete recordings with retention control |
| /v2/numbers/search | GET | Search live DID inventory by country, city and capability |
| call.answered / call.completed | webhook | Signed lifecycle events with full call metadata |
| recording.ready | webhook | Fired when a recording is durable and downloadable |
| dtmf.received | webhook | Real-time digit capture during an active call |
OpenAPI 3.1 specification and Postman collection are published and version-controlled alongside the API.
Usage-based API pricing
You pay for calls and features consumed. There is no platform fee, no seat licence and no charge for the sandbox.
- Unlimited test calls
- Reserved test numbers
- Full API surface access
- Webhook inspector
- All SDKs and documentation
- Community support
- Inbound from $0.0055 per minute
- Recording $0.0015 per minute
- Speech recognition $0.02 per minute
- Text-to-speech $0.006 per 1k characters
- Media streaming included
- 24/7 NOC support
- Negotiated per-minute rates
- Dedicated capacity and rate limits
- Private interconnect option
- Custom SLA with financial remedies
- Early access to new endpoints
- Named solutions engineer
All prices are illustrative USD list rates excluding local taxes and regulatory fees. Committed-volume and multi-year agreements are quoted individually — request a tailored quote.
Voice API questions, answered plainly
Still unsure? Our voice architects answer technical questions directly — no gatekeeping through sales.
Yes, and it does not expire. You get API credentials, reserved test numbers and unlimited test calls between them, with the full API surface available — not a cut-down version.
The only thing sandbox will not do is place calls to real PSTN numbers, since that costs us money. Everything else behaves identically to production.
The API is versioned in the path. Breaking changes only ship in a new major version, and the previous version is supported for a minimum of twelve months after its successor is released.
Deprecations are announced by email to every account using the affected endpoint, flagged in response headers, and accompanied by a written migration guide with code examples. We have never silently changed behaviour in a live version.
Default production limits are 100 requests per second per account with a burst allowance of 300, and 500 concurrent calls. Every response carries standard rate-limit headers so you can back off cleanly rather than guessing.
Limits are raised on request with a brief description of your traffic pattern — usually approved within a business day, and free of charge.
Yes. The media streaming endpoint opens a bidirectional WebSocket carrying raw audio for either or both legs, and you can inject synthesised audio back into the call on the same socket.
Added latency is typically under 100 milliseconds, which is what makes conversational voice agents feasible rather than merely demonstrable. Reference implementations for common speech providers are published in the docs.
Yes. You can originate and terminate over SIP alongside the HTTP API, and native SIP REFER handling means calls hand off cleanly to on-premise systems.
This matters for enterprises with an existing PBX estate: the API can add programmable capability without requiring you to move core telephony first.
Sandbox accounts get community support and full documentation. Production accounts get 24/7 NOC access, and Scale customers get a named solutions engineer.
For launches, any production customer can request a technical contact on shift during their go-live window at no cost — tell us the date and we will staff for it.
Start building in the next ten minutes
Self-service sandbox credentials, reserved test numbers and unlimited test calls — no card, no sales call, no trial countdown. Talk to us only when you are ready for production traffic.
30-day trial · 500 free minutes · No credit card required