Dial

Definition

The Dial verb connects the calling party with the dialed party. Once the call is set up, the participants can start talking.

When the call is ended successfully, or the dialed number is not reached, or the call is terminated, CarrierX will make a POST or GET request to the action URL.

warning

If the Dial verb specifies an action URL, FlexML instructions after the Dial will not be executed. Control is tranferred to the Response returned from the action URL.

If no action URL is set, the call flow will move to the next verbs in the FlexML instructions.

You can also place a phone number inside the Number noun. This is done to allow specifying the additional parameters (e.g., sendDigits) when dialing a phone number. Refer to the this section for more information about the Number noun.

Additionally, you can use the Number noun for the simultaneous ring to several numbers feature.

When to Use

Use Dial when you want to connect the current caller to another phone number, SIP endpoint, or multiple phone numbers or SIP endpoints at once. Dial controls what happens if the call is answered, not answered, or terminated.

info

Dial can record a call, stream a call, or do both at the same time.

• To record a call, use the record attributes.
• To stream a call, use the stream attributes.

tip

• Dial works with regular phone numbers. SIP URI dialing is supported only when routing through a configured trunk group (via a trunk group SID). Direct SIP-to-IP dialing (e.g., to arbitrary IP addresses) is not supported for security reasons.
• Always handle your action URL to manage what happens next when a Dial ends, so you can gracefully redirect, play a message, or log the result.

Supported Attributes

These attributes can be used to modify the Dial verb. They are inserted as name-value pairs in the start-tag.

Attribute	Data Type	Description
action	string	The link to another set of FlexML instructions. A request will be made to this URL when the call is ended or if the dialed party is not available. If no action URL is set, the rest of the verbs in the current FlexML instructions will be executed. The URL can be absolute or relative. When triggered, the callback includes standard call parameters such as CallStatus as well as detailed call timing fields with millisecond precision: StartTime — when the Dial began EndTime — when the Dial ended (null if ongoing) AnswerTime — when the call was answered CallDuration — total talk time for the call DialStartTime, DialEndTime, DialAnswerTime, DialCallDuration — timings specific to the last Dial operation. For more details about timing parameters included in callbacks, see Callback Timing Parameters.
background	boolean	When set to true, enables background dial mode. In this mode, audio from the current call is forked outbound to the specified number over SIP. FlexML execution continues immediately; the Dial verb does not block further instructions. Only the first Number is used. Additional numbers are ignored. Audio flow is one-way only: the dialed party receives the caller's audio, but no audio is sent back into the main call. Useful as an alternative to the Stream verb when you prefer SIP instead of WebSockets. The default value is false. For details, see Background Dial.
callerId	string	The phone number that will appear on the caller ID of the called party. This phone number must be on the account, or have custom permission set up. Also supported for background dial calls.
callerName	string	The calling party CNAME that is used in the from field.
confirmKey	string	The key that the called party must press to start the phone talk. Values accepted in this field are: 0-9, #, and *.
confirmSound	string	The URL of the audio that contains the instructions for the key from the confirmKey field. It informs the called party which key they must press to start the talk.
containerSid	string	The secure ID of the container to which the recording will be written.
hangupOnStar	boolean	If this value is set to true, pressing * will hang up the call with the dialed number, and continue with the next FlexML instructions. Until * is pressed, the rest of the FlexML instructions will not be executed.
method	string	The type of request made to the action URL. Values accepted in this field are POST or GET. The default value is POST.
record	string	Determines whether or not to record the outbound call until hangup. Values accepted in this field are: do-not-record for the call not to be recorded (the same effect as when the record attribute is not specified). record-from-answer for recording to start from the moment the call is answered. record-from-ringing for the recording to start from the moment the ringing begins. The default value is do-not-record.
recordDirection	string	The direction of the call that will be recorded. Values accepted in this field are: any to allow recording of both called and calling parties. in to allow recording of the calling party only. out to allow recording of the called party only. The default value is any.
recordingStatusCallback	string	A callback is sent to this URL after the recording has ended.
recordingStatusCallbackEvent	string	The comma-separated list of events to be notified about. Options are completed, to indicate that the recording is successful, and failed, to indicate that the recording has failed. The default value is "completed,failed".
recordingStatusCallbackMethod	string	The method, either POST or GET that the callback to recordingStatusCallback is sent. The default value is POST.
ringing	string	Phone ringing management. Values accepted in this field are: true: return a 180 Ringing informational response—ringing tone must be generated by the called party. music: play default hold or ringing music (after answering the call for compatibility). false: play silence instead of ringing, but still answers the call (to remain compatible with legacy flows). early_media_music: attempt to play music using an early-media response, without answering the call. early_media_silence: attempt to play silence using early media, without answering the call — useful for silent connects where media starts before bridging. Default behavior remains true if not specified.
stream	string	Determines whether or not to stream the outbound call audio. Values accepted in this field are: do-not-stream for the call not to be streamed (the same effect as when the stream attribute is not specified). stream-from-answer for streaming to start from the moment the call is answered. stream-from-ringing for streaming to start from the moment the ringing begins. The default value is do-not-stream.
streamEncoding	string	Determines the audio encoding format for the stream. Supported values: audio/x-mulaw - default encoding. audio/L16 - linear 16-bit PCM (Pulse-Code Modulation). Note: If the call uses an HD (wideband) codec, using audio/L16 can preserve higher audio quality. For non-HD codecs, audio/L16 works but doesn't improve quality — it only increases bandwidth.
streamMaxLength	integer	The maximum length of the stream, in seconds.
streamName	string	A unique identifier for the stream.
streamTimestampStart	string	Defines how the timestamp attribute of the streamed media is represented. Values accepted in this field are: relative (default) absolute
streamTrack	string	The audio track(s) to stream. Values accepted in this field are: inbound outbound both - stream both inbound and outbound tracks individually. mixed - stream inbound and outbound as a single, mixed track.
streamUrl	string	The URL of the WebSocket server. Both ws and wss protocols are supported.
timeLimit	integer	The maximum number of seconds that the call will last. The default value is 0, meaning that the call time is not limited. Also applies to background dial calls.
timeout	integer	The number of seconds to wait for the destination number to answer. After the call times out, a request is made to the action URL. The default value is 30.
trim	string	Determines whether or not silence is removed from the beginning or ending of a recording. To trim the silence in the recording, set this field to trim-silence. The default value is do-not-trim. Also applies to background dial calls.

note

• recordDirection, recordingStatusCallback, recordingStatusCallbackEvent, recordingStatusCallbackMethod, and trim are applicable only when record is set to record-from-answer or record-from-ringing.

• streamTrack, streamMaxLength, streamName, streamTimestampStart, and streamUrl are applicable only when stream is set to stream-from-answer or stream-from-ringing.

Examples

To simply dial the number, include it directly inside the Dial verb without using the Number noun. In this case, FlexML instructions cannot include any additional parameters for the dialed number. In this example, the phone number inside the Dial tags will be dialed. The call total time will be limited to 15 seconds. The called party will be able to hang up by pressing the * key. If an action URL is added to the Dial verb as an attribute, a request will be made to that URL if the phone number cannot be reached, or when the call ends.

<Response>
  <Dial timeLimit="15" hangupOnStar="true">15162065340</Dial>
</Response>

Dial with streaming

In this example, the Dial verb streams the call audio to a WebSocket server after the call is answered.

<Response>
  <Dial stream="stream-from-answer" streamUrl="wss://example.com/stream" streamName="myStream" streamTrack="both"></Dial>
</Response>

Dial with streamEncoding

In this example, the Dial verb specifies the streamEncoding attribute to ensure audio is streamed using linear 16-bit encoding (audio/L16), which preserves HD quality if available.

<Response>
  <Dial stream="stream-from-answer" streamEncoding="audio/L16" callerId="+14155550100" streamUrl="wss://example.com/stream" streamName="myStream" streamTrack="both">
    +14155550101
  </Dial>
</Response>

Number Noun

The Number noun can be nested inside the Dial verb. It specifies the DID number to be dialed. It supports dialing to standard phone numbers and can optionally specify a trunk group SID for forced routing.

Forced Routing to a Trunk Group

You can append an @ symbol followed by a valid trunk group SID to force the call to be routed through that specific trunk group.

Example syntax:

<Dial>
  <Number>15162065340@b435ee72-36f4-48f8-9204-ae62a3e73119</Number>
</Dial>

Direct SIP URI dialing to arbitrary IP addresses (for example, 12342342345@123.123.123.123) is not supported due to firewall restrictions. Instead, you can append an @ symbol followed by a valid trunk group SID to force routing through a specific trunk group.

note

When using background="true" in the Dial verb, only the first Number noun will be used - any additional numbers will be ignored. Simultaneous ringing is not supported in background dial mode.

Supported Attributes

Attribute	Data Type	Description
sendDigits	string	The DTMF tones that will be played when the call is answered. This is especially useful when dialing a phone number with an extension. Values accepted in this field are: 0-9, #, *, ,, and the combinations of these characters. Each comma character defines a 0.5-second delay.

CarrierX allows dialing to several numbers at once using the Number noun. Refer to the Simultaneous Ring to Several Numbers section of the Dial verb for more information about this.

Query Arguments

You can also send custom SIP headers using query arguments. Use the ? character after the phone number to add these headers. Multiple headers are joined with the XML-encoded & character.

In this example, two X- headers will be sent when dialing the phone number specified in the Number noun nested inside the Dial verb.

<Response>
  <Dial>
    <Number>15162065340?X-CustomHeader1=ABC&amp;X-CustomHeader2=DEF</Number>
  </Dial>
</Response>

The additional query arguments can be used to send custom headers inside the Number noun. Use the ? character after the phone number to add these headers. Several headers are joined with the XML-encoded & character.

info

Only X- headers are accepted as query arguments, any other headers or strings will be filtered out and not sent.

note

If you choose to send custom headers when dialing a phone number, you will need to additionally set the relay_sip_headers parameter of the Trunk object associated with your FlexML endpoint to successfully pass these headers to the destination.

Examples

DTMF with a Delay

In this example, once the call is answered, the system waits two seconds (four commas equal to 0.5-second each) before playing the 1928 DTMF tones.

<Response>
  <Dial>
    <Number sendDigits=",,,,1928">15162065340</Number>
  </Dial>
</Response>

Forced Routing to a Trunk Group

In this example, the call to the number will be routed through a specific trunk group.

<Response>
  <Dial>
    <Number>15162065340@22a24d1b-e9ec-465d-9ff9-31b67817bb58</Number>
  </Dial>
</Response>

Simultaneous Ring to Several Numbers

CarrierX supports dialing to several numbers at once. You need to specify the numbers in the Number noun tags inside the Dial verb. The first person to answer will be connected to the calling party. The calls to the other numbers will be hung up.

note

CarrierX supports up to 10 concurrent rings using the FlexML Dial verb.

tip

You can see an example of the code that uses the simultaneous rings feature in our sample application, which is a part of the No-code CarrierX Apps collection.

Examples

Basic Simultaneous Ring

In this example, the phone numbers in the Number tags will be dialed all at the same time. The first person to answer is connected to the calling party.

<Response>
  <Dial>
    <Number>12345645585</Number>
    <Number>96846465465</Number>
    <Number>98133217461</Number>
  </Dial>
</Response>

Simultaneous Ring with ConfirmKey and Recording

In this example, multiple phone numbers inside the Number tags will be dialed simultaneously. The first person to answer will be connected to the calling party, and they must press 1 (as instructed by the audio at confirmSound) to start the conversation. The call will be recorded starting from when it is answered, and the total call time is limited to 30 seconds.

<Response>
  <Dial confirmKey="1" confirmSound="https://example.com/instructions.mp3" record="record-from-answer" timeLimit="30">
    <Number>1234567890</Number>
    <Number>0987654321</Number>
  </Dial>
</Response>

Background Dial

Background dial lets you fork live call audio to a SIP connection, enabling use cases such as real-time transcription, voice recognition, and other audio processing applications.

Normally, the Dial verb is blocking: when executed, FlexML waits until the dialed call ends (either answered and completed, not answered, or failed) before continuing to the next verbs. Any verbs listed after Dial do not run until the dial completes.

When you set background="true", Dial becomes non-blocking. The outbound leg is initiated in parallel, audio from both directions is forked to your application, and FlexML execution continues immediately with the next verbs in your script. The flow does not wait for the background call to connect or finish.

Key differences from regular Dial

• Regular Dial: blocks FlexML until the dial attempt completes.
• Background Dial: non-blocking; execution continues immediately. Audio is forked outbound but not bridged.

Other characteristics

• Only the first Number specified will be dialed; additional numbers are ignored.
• Maximum of 3 concurrent background dial sessions are allowed (same as the background recording limit).

Example

<Response>
  <Dial background="true" timeout="60">
    <Number>18083842000</Number>
  </Dial>
  <Say>This message plays immediately while the background dial is active.</Say>
</Response>

Here, the caller's audio is forked to 18083842000 over SIP, while the call flow continues to play a message.

Callback Data

In addition to the standard callback fields the following fields are returned in the callback request body when the Dial verb completes:

• If an action URL is specified, these fields are sent immediately in the action request.
• If no action URL is specified, FlexML continues with the next instruction, but the Dial-related fields are still included in all subsequent callbacks (including the final status callback).

Attribute	Data Type	Description
AnswerTime	string	Timestamp when the call was fully connected and both parties could start talking.
CallDuration	integer	Duration in seconds from when the call was answered until it ended. Will be null if the call is still ongoing when the action URL is requested.
DialAnswerTime	string	Timestamp when the most recent Dial verb was fully connected.
DialCallDuration	integer	Duration in seconds of the talk time for the most recent Dial verb.
DialEndTime	string	Timestamp when the most recent Dial verb execution ended.
DialStartTime	string	Timestamp when the most recent Dial verb execution started.
EndTime	string	Timestamp (with millisecond precision) when the overall call ended. If the action callback is triggered before the call ends, this value will be null.
StartTime	string	Timestamzp (with millisecond precision) indicating when the FlexML flow initially started executing the Dial verb.

If multiple Dial verbs are executed in the same call flow, the Dial* fields always reflect the most recent Dial verb.

If the callback occurs while the call is still active (e.g., an in-progress action callback), EndTime, CallDuration, and their Dial-specific equivalents may be null.

A typical callback returned when the Dial verb ends might look like this:

{
  "AccountSid": "1234567890abcdef",
  "AnswerTime": 1723480521.123,
  "ApiVersion": "2.0",
  "CallDuration": 17,
  "CallerName": "Lucas",
  "CallSid": "6aa6217e281e4872af59b79f813fdaf7",
  "CallStatus": "completed",
  "DialAnswerTime": 1723480521.123,
  "DialCallDuration": 17,
  "DialEndTime": 1723480538,
  "DialStartTime": 1723480520,
  "Direction": "inbound",
  "EndTime": 1723480538,
  "From": "15627317020"
  "OriginalFrom": "15627317020",
  "OriginalTo": "15162065599",
  "PartnerSid": "32c51df0b2f54845aef695fcd07f3994",
  "RequestUrl": "http://10.222.2.129:5005/stream",
  "StartTime": 1723480518,
  "To": "15162065599",
}

Early-Media Support for Ringing

With early_media_music and early_media_silence, you can now initiate media playback before the call is answered, using early media. This means the caller hears audio (music or silence), but the call leg to the callee is not yet answered, preserving signaling and avoiding billing triggers until answer.

These modes can reduce “click” artifacts or abrupt transitions when connecting — particularly useful for integrations that require a smooth audio handoff or want to defer formal answer until later in the call flow.

Example

<Response>
  <Dial ringing="early_media_music" timeLimit="60">
    1234567890
  </Dial>
</Response>

This example will play early-media music to the caller without answering the call while dialing the destination. If the call is answered, the flow proceeds normally; if not, you can handle timeout or fallback accordingly.

StreamParameter

You can use the StreamParameter noun inside the Dial verb to pass custom key/value pairs along with the dialed call. This works in the same way as the Parameter noun that is used with Stream. Each StreamParameter requires a name and a value.

For more details on supported attributes, see Parameter.

Example

<Response>
  <Dial>
    <StreamParameter name="FirstName" value="John"/>
    <StreamParameter name="LastName" value="Cleese"/>
      15551234567
  </Dial>
</Response>

Common Pitfalls

• If action is set, any verbs after Dial won't run. The Response from the action URL is executed.
• To pass SIP headers or custom headers, use ?X-Header=Value in the Number noun.
• Remember to set relay_sip_headers in your Trunk if you pass custom SIP headers.

Nesting Rules

• You can nest the Number noun within the Dial verb.
• You can nest StreamParameter within Dial to pass custom key/value pairs for the dial stream. StreamParameter behaves exactly like the Parameter noun (same supported attributes and restrictions), but is scoped to Dial.
• You cannot nest the Dial verb within other FlexML verbs.

Learn more

• Connect — Start a bidirectional stream with a WebSocket.
• Record — Record the entire call or only part of it.
• Stream — Stream audio to a WebSocket.
• Start / Stop — Control unidirectional streams.