Wirexa API — Voice Call API Reference

Wirexa API is a CPaaS (Communications Platform as a Service) that lets you make and receive voice calls programmatically. It is compatible with Twilio and SignalWire SDKs — you can migrate existing code by changing the base URL.

Base URL for all API requests:

https://api.wirexaapi.com

Twilio / SignalWire compatible: If you already use the SignalWire Python SDK, just replace signalwire_space_url with wirexaapi.com and the rest of your code stays the same.

Quick Start

Make your first call in under 2 minutes.

1. Install the SDK

bash

pip install wirexaapi

2. Make a call

python

from wirexaapi import Client

client = Client(
    "your_project_id",
    "your_api_token",
    signalwire_space_url="wirexaapi.com"
)

call = client.calls.create(
    to="+15551234567",
    from_="+15559876543",
    url="https://your-backend.com/voice",
    status_callback="https://your-backend.com/status",
    record=True,
)

print(call.sid, call.status)
# CAxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx  queued

3. Return LAML from your backend

When the call connects, Wirexa API sends a POST request to your url. Your backend must respond with LAML/XML:

python (Flask)

from flask import Response

def voice():
    xml = """<Response>
  <Say language="en-US">Hello! Thanks for calling.</Say>
  <Gather numDigits="1" timeout="7"
          action="https://your-backend.com/gather">
    <Say>Press 1 for sales, 2 for support.</Say>
  </Gather>
  <Hangup/>
</Response>"""
    return Response(xml, mimetype="text/xml")

Authentication

All REST API requests use HTTP Basic Authentication. Use your Project ID as the username and your API Token as the password.

curl

curl -X POST https://api.wirexaapi.com/api/laml/2010-04-01/Accounts/{ProjectID}/Calls.json \
  -u "{ProjectID}:{ApiToken}" \
  -d "To=+15551234567" \
  -d "From=+15559876543" \
  -d "Url=https://your-backend.com/voice"

You can find your credentials in the dashboard under Project Settings.

Python SDK

The wirexaapi package is a drop-in replacement for the SignalWire Python SDK.

bash

pip install wirexaapi

Migrating from SignalWire? Replace from signalwire.rest import Client with from wirexaapi import Client and set signalwire_space_url="wirexaapi.com". No other changes needed.

Create Call

Initiates an outbound call to the specified number.

POST /api/laml/2010-04-01/Accounts/{AccountSid}/Calls.json

Request Parameters

Parameter	Required	Description
`To`	Required	Destination phone number in E.164 format (e.g. `+15551234567`)
`From`	Required	Caller ID number assigned to your account
`Url`	Required	URL that returns LAML/XML to control the call flow
`Method`	Optional	HTTP method to request `Url`. Default: `POST`
`Timeout`	Optional	Seconds to wait for answer before giving up. Default: `28`
`Record`	Optional	`true` to record the entire call. Default: `false`
`StatusCallback`	Optional	URL to receive call status updates (answered, completed, etc.)
`StatusCallbackEvent`	Optional	Space-separated events: `initiated ringing answered completed`
`RecordingStatusCallback`	Optional	URL to receive recording completion notification
`MachineDetection`	Optional	`Enable` or `DetectMessageEnd` — enables answering machine detection
`AsyncAmd`	Optional	`true` to detect AMD asynchronously while the call proceeds
`AsyncAmdStatusCallback`	Optional	URL to receive async AMD result

Response (201 Created)

json

{
  "sid": "CAxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
  "account_sid": "your_project_id",
  "to": "+15551234567",
  "from": "+15559876543",
  "status": "queued",
  "direction": "outbound-api",
  "date_created": "Mon, 16 Mar 2026 10:00:00 +0000"
}

Python SDK

python

call = client.calls.create(
    to="+15551234567",
    from_="+15559876543",
    url="https://your-backend.com/voice",
    status_callback="https://your-backend.com/status",
    record=True,
    machine_detection="Enable",
    async_amd=True,
    async_amd_status_callback="https://your-backend.com/amd",
)

Get Call

Retrieve the current status and details of a call.

GET /api/laml/2010-04-01/Accounts/{AccountSid}/Calls/{CallSid}.json

python

call = client.calls("CAxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx").fetch()
print(call.status)    # queued | ringing | in-progress | completed | failed | no-answer | busy
print(call.duration)  # duration in seconds (available after completion)

Modify Call

Update an active call — redirect it to a new URL or hang it up.

POST /api/laml/2010-04-01/Accounts/{AccountSid}/Calls/{CallSid}.json

Parameter	Description
`Status`	Set to `completed` to hang up the call
`Url`	Redirect the call to a new LAML URL

python

# Hang up
client.calls("CAxxxx").update(status="completed")

# Redirect to new flow
client.calls("CAxxxx").update(url="https://your-backend.com/new-flow")

List Calls

GET /api/laml/2010-04-01/Accounts/{AccountSid}/Calls.json

python

calls = client.calls.list(limit=50)
for call in calls:
    print(call.sid, call.status, call.to)

LAML / XML — Introduction

LAML (Language Markup Language) is an XML-based language that controls call flows. When Wirexa API connects a call, it fetches your Url and executes the verbs in order.

<Response>
  <Say>Hello world.</Say>
  <Hangup/>
</Response>

All responses must be valid XML with Content-Type: text/xml.

Available Verbs

<Say>

Convert text to speech using the configured TTS provider.

<Play>

Play an audio file (MP3 or WAV) from a public URL.

Collect DTMF digits from the caller. Can nest Say/Play.

Record the caller's voice. Saves as WAV and MP3.

End the call immediately.

Transfer the call to another number or SIP endpoint.

<Say>

Converts text to speech and plays it to the caller.

Attribute	Default	Description
`language`	`en-US`	BCP-47 language tag (e.g. `es-CO`, `en-GB`, `pt-BR`)
`voice`	`woman`	Voice to use. See TTS providers for supported values.
`loop`	`1`	Number of times to repeat. `0` = infinite.

<Response>
  <Say language="en-US" voice="polly.Joanna">
    Your appointment is confirmed for tomorrow at 3 PM.
  </Say>
</Response>

<Play>

Plays an audio file from a public URL.

<Response>
  <Play>https://your-cdn.com/audio/greeting.mp3</Play>
</Response>

<Gather>

Waits for the caller to press digits (DTMF). Posts the collected input to action.

Attribute	Default	Description
`action`	Current URL	URL to POST the collected digits to
`numDigits`	unlimited	Stop after this many digits
`timeout`	`5`	Seconds of silence before timeout
`finishOnKey`	`#`	Key that ends input collection
`method`	`POST`	HTTP method for `action`

<Response>
  <Gather numDigits="1" timeout="7"
          action="https://your-backend.com/gather">
    <Say>Press 1 for sales, 2 for support, 3 to repeat.</Say>
  </Gather>
  <Say>We did not receive your input. Goodbye.</Say>
  <Hangup/>
</Response>

Your action endpoint receives Digits in the POST body:

python (Flask)

def gather():
    digits = request.form.get("Digits")
    if digits == "1":
        xml = "<Response><Say>Transferring to sales.</Say></Response>"
    elif digits == "2":
        xml = "<Response><Say>Transferring to support.</Say></Response>"
    else:
        xml = "<Response><Redirect/></Response>"  # repeat
    return Response(xml, mimetype="text/xml")

<Record>

Records the caller's voice. Sends a webhook when recording is complete.

Attribute	Default	Description
`action`		URL to redirect after recording stops
`maxLength`	`3600`	Maximum recording duration in seconds
`timeout`	`5`	Seconds of silence before auto-stop
`finishOnKey`	`#`	Key that stops recording
`recordingStatusCallback`		URL to POST when recording file is ready

<Response>
  <Say>Please leave your message after the tone.</Say>
  <Record maxLength="60" timeout="3"
          recordingStatusCallback="https://your-backend.com/recording"/>
  <Say>Thank you. Goodbye.</Say>
  <Hangup/>
</Response>

<Hangup>

Ends the call immediately. No attributes required.

<Response><Hangup/></Response>

<Transfer> / <Dial>

Transfers the call to another phone number.

<Response>
  <Say>Transferring your call now.</Say>
  <Transfer>+15559998888</Transfer>
</Response>

Answering Machine Detection (AMD)

AMD automatically detects whether a human or an answering machine answered the call. Use it to skip voicemails or play a different message.

Async AMD (recommended)

The call proceeds immediately while AMD runs in the background. The result is sent to AsyncAmdStatusCallback.

python

call = client.calls.create(
    to="+15551234567",
    from_="+15559876543",
    url="https://your-backend.com/voice",
    machine_detection="Enable",
    async_amd=True,
    async_amd_status_callback="https://your-backend.com/amd",
)

Your AMD callback receives:

AnsweredBy

human or machine_start or machine_end_beep or fax

CallSid

The call SID

MachineDetectionDuration

Milliseconds taken to detect

python (Flask — AMD callback)

def amd_callback():
    answered_by = request.form.get("AnsweredBy")
    call_sid    = request.form.get("CallSid")

    if answered_by == "human":
        xml = "<Response><Say>Hello! This is a reminder for your appointment.</Say></Response>"
    else:
        # Machine — hang up or leave a different message
        xml = "<Response><Hangup/></Response>"

    return Response(xml, mimetype="text/xml")

Recordings

Call recordings are stored server-side and accessible via public URLs. Both WAV and MP3 formats are available.

Format	URL
`WAV`	`https://api.wirexaapi.com/recordings/{RecordingSid}.wav`
`MP3`	`https://api.wirexaapi.com/recordings/{RecordingSid}.mp3`

The RecordingSid is delivered via the RecordingStatusCallback webhook once the file is ready.

Webhooks

Wirexa API sends two separate POST callbacks at the end of each call.

Call Status Callback

Sent to StatusCallback when the call status changes.

CallSid

Unique call identifier

CallStatus

Destination number

From

Caller ID number

Direction

outbound-api | inbound

CallDuration

Duration in seconds

AnsweredBy

AMD result if enabled

Recording Status Callback

Sent to RecordingStatusCallback when the recording file is ready.

RecordingSid

Unique recording identifier

RecordingUrl

Base URL for the recording (append .wav or .mp3)

RecordingStatus

completed

RecordingDuration

Duration in seconds

CallSid

Associated call SID

Text-to-Speech Providers

Specify the provider and voice in the voice attribute of <Say>.

Voice value	Provider	Example
`polly.Joanna`	Amazon Polly	`<Say voice="polly.Joanna">`
`polly.Miguel`	Amazon Polly (es)	Spanish male voice
`google.en-US-Wavenet-F`	Google Cloud TTS	WaveNet female
`google.es-US-Wavenet-B`	Google Cloud TTS	Spanish WaveNet
`eleven.Rachel`	ElevenLabs	Hyper-realistic voice
`azure.en-US-JennyNeural`	Azure Cognitive	Neural voice
`woman`	Auto (best available)	Automatic fallback chain
`man`	Auto (best available)	Automatic fallback chain

If a provider fails, the engine automatically falls back to the next available provider — ensuring your call always plays audio.