Documented WebSocket (wss://) voice-streaming API. When a call reaches a Voicebot or Stream applet, Exotel opens a secure WebSocket to your endpoint and streams base64 linear-PCM audio in ~100 ms frames (connected/start/media/dtmf/mark/stop); the Voicebot applet is bidirectional so your server sends media/mark/clear back to speak to the caller and barge in. Exotel is the WebSocket client; you host the endpoint.
Every API here is available over the APIs.io API and to AI agents over MCP.
One button, every client — Claude, Cursor, VS Code and the rest.
https://apis.io/mcp
find_apisBrowse and filter every API in the catalog.get_api_artifactsOne API's artifacts, grouped by type.get_openapiThe primary OpenAPI for this API.find_similar_apisAPIs that look like this one.apis_io_searchSTART HERE — APIs, providers and tags for one query, each with its total.resolveTurn a domain, URL or GitHub org into the provider it belongs to.find_cohortsEvery scored population of providers in the catalog.curl "https://apis.io/api/v1/apis/exotel:exotel-agentstream-voice-streaming-api"curl "https://apis.io/api/v1/apis?limit=25"Calls without X-API-Key need no key. Calls with it need a paid plan — get a key.
Free tier, no form to fill in. Signing in shares your email address with us — we store it to create your key and to recognise you if you sign in with another provider. See our Privacy Policy and Terms.
A second provider on the same verified email joins the account you already have.
asyncapi: 2.6.0
info:
title: Exotel AgentStream Voice Streaming API
version: '1.0'
description: >-
Exotel AgentStream (also surfaced through the Voicebot and Stream
applets) is a documented public WebSocket API for real-time voice media
streaming during a call. When a call hits a Voicebot/Stream applet in an
Exotel call flow, Exotel opens a secure WebSocket (`wss://`) connection
to the endpoint you configure and streams the caller's audio as
base64-encoded linear PCM in ~100 ms frames. With the Voicebot applet the
stream is BIDIRECTIONAL: your server can send audio (and `mark`/`clear`
control events) back over the same socket to speak to the caller and to
barge in. The Stream applet is unidirectional (caller audio to your
server only).
CONNECTION MODEL: Exotel is the WebSocket CLIENT; your bot/server hosts
the `wss://` endpoint. This document therefore models YOUR server as the
AsyncAPI "server" that Exotel connects to. Replace the host below with
your own endpoint.
GROUNDING: Event names and JSON message shapes (connected, start, media,
dtmf, mark, stop inbound; media, mark, clear outbound) and the audio
format are taken from Exotel's AgentStream developer guide. Field-level
schemas are modeled from the documented example messages.
contact:
name: Exotel Developer Support
url: https://developer.exotel.com/docs/agentstream/developer-guide
x-transport-notes: >-
Real bidirectional WebSocket (wss://) with a documented message protocol.
This is NOT HTTP SSE. Exotel initiates the connection to your endpoint;
the URL, path, and query string (which carry custom_parameters) are
defined by you when you configure the Voicebot/Stream applet.
servers:
yourEndpoint:
url: your-bot-host.example.com
protocol: wss
description: >-
Your WebSocket endpoint that Exotel connects to. You supply a wss://
(or https:// upgraded) URL in the Voicebot/Stream applet configuration;
any path and query parameters you set are echoed back as
custom_parameters in the start event.
defaultContentType: application/json
channels:
/:
description: >-
The single bidirectional media stream socket for one call. Exotel sends
connected/start/media/dtmf/mark/stop; your server (Voicebot applet only)
may send media/mark/clear.
subscribe:
operationId: receiveFromExotel
summary: Messages Exotel sends to your server.
message:
oneOf:
- $ref: '#/components/messages/Connected'
- $ref: '#/components/messages/Start'
- $ref: '#/components/messages/MediaIn'
- $ref: '#/components/messages/Dtmf'
- $ref: '#/components/messages/MarkIn'
- $ref: '#/components/messages/Stop'
publish:
operationId: sendToExotel
summary: Messages your server sends back to Exotel (Voicebot applet only).
message:
oneOf:
- $ref: '#/components/messages/MediaOut'
- $ref: '#/components/messages/MarkOut'
- $ref: '#/components/messages/Clear'
components:
messages:
Connected:
name: connected
title: Connected
summary: First frame after the socket opens.
payload:
type: object
properties:
event:
type: string
const: connected
Start:
name: start
title: Start
summary: Sent once at stream start with call and media metadata.
payload:
type: object
properties:
event:
type: string
const: start
sequence_number:
type: string
stream_sid:
type: string
start:
type: object
properties:
stream_sid:
type: string
call_sid:
type: string
account_sid:
type: string
from:
type: string
to:
type: string
custom_parameters:
type: object
additionalProperties: true
description: Query parameters you set on your applet URL.
media_format:
type: object
properties:
encoding:
type: string
example: audio/x-raw
sample_rate:
type: string
enum: ['8000', '16000', '24000']
bit_rate:
type: string
example: '16'
MediaIn:
name: media
title: Media (inbound)
summary: A frame of caller audio (base64 linear PCM).
payload:
type: object
properties:
event:
type: string
const: media
sequence_number:
type: string
stream_sid:
type: string
media:
type: object
properties:
chunk:
type: string
timestamp:
type: string
payload:
type: string
description: Base64-encoded 16-bit mono linear PCM.
Dtmf:
name: dtmf
title: DTMF (Voicebot only)
summary: A DTMF keypress detected during the call.
payload:
type: object
properties:
event:
type: string
const: dtmf
sequence_number:
type: string
stream_sid:
type: string
dtmf:
type: object
properties:
digit:
type: string
duration:
type: string
MarkIn:
name: mark
title: Mark (inbound, Voicebot only)
summary: Confirms previously sent audio has finished playing.
payload:
type: object
properties:
event:
type: string
const: mark
sequence_number:
type: string
stream_sid:
type: string
mark:
type: object
properties:
name:
type: string
Stop:
name: stop
title: Stop
summary: Sent when the stream ends.
payload:
type: object
properties:
event:
type: string
const: stop
sequence_number:
type: string
stream_sid:
type: string
stop:
type: object
properties:
call_sid:
type: string
account_sid:
type: string
reason:
type: string
enum: [stopped, callended]
MediaOut:
name: media
title: Media (outbound, Voicebot only)
summary: A frame of audio to play to the caller (base64 PCM).
payload:
type: object
properties:
event:
type: string
const: media
stream_sid:
type: string
media:
type: object
properties:
payload:
type: string
description: >-
Base64 PCM/PCMU in ~100 ms frames; multiples of 320 bytes,
up to ~100 KB per frame.
MarkOut:
name: mark
title: Mark (outbound, Voicebot only)
summary: Label a point in your outbound audio stream.
payload:
type: object
properties:
event:
type: string
const: mark
stream_sid:
type: string
mark:
type: object
properties:
name:
type: string
Clear:
name: clear
title: Clear (outbound, Voicebot only)
summary: Barge-in - discard queued outbound audio.
payload:
type: object
properties:
event:
type: string
const: clear
stream_sid:
type: string