> Fetch clean Markdown by appending `.md` to any page URL under https://signalwire.com/docs or requesting it with the HTTP header `Accept: text/markdown`. The root index at https://signalwire.com/docs/llms.txt lists the available documentation indexes.

# Gathering user input

> Use cXML to prompt and receive user input.

This article explains how to gather input from users using cXML Scripts.

In general, there are three kinds of input that you could want to gather: keypad
input, text input via speech recognition, and audio recordings. In this article
we are going to focus on keypad input and speech recognition.

## cXML for Gathering Input

We are going to define the call handling instructions in a cXML Script hosted on SignalWire.

#### New Dashboard

To create a cXML script, navigate to the **Resources** tab in your [SignalWire Dashboard](https://my.signalwire.com), click **+ Add New**, and select **cXML Script**.

#### Legacy Dashboard

If you're on the Legacy UI, go to the cXML/LaML section in your SignalWire Space, then click on Bins, and create a new script.

#### Which Dashboard version am I using?

**New Dashboard**: You have a **Resources** tab in the left sidebar.

**Legacy Dashboard**: You have separate tabs for SIP, LaML, Relay, etc.

Paste the following XML content in your new cXML Script:

```xml
<?xml version="1.0" encoding="UTF-8"?>
<Response>
  <Gather
    input="speech dtmf"
    action="https://example.com/my-webhook"
    timeout="5"
    numDigits="5"
    hints="one two three four five six seven eight nine">
    <Say>
      Welcome! Please enter or say your account number.
    </Say>
  </Gather>

  <Say>No input detected.</Say>
  <Hangup />
</Response>
```

We used the [`<Gather>`](/docs/compatibility-api/cxml/reference/voice/gather) verb to gather input as soon as the call starts. We set five attributes:

* `input`, which specifies which kind of input we want to gather (in our case, both speech and DTMF keypad input);
* `action`, which specifies the URL to fetch when the input has been collected (the URL should return a new cXML document to execute);
* `timeout`, i.e., the number of seconds of silence or inaction that denote the end of caller input;
* `numDigits`, which indicates the number of digits to collect via DTMF keypad input; and
* `hints`, an optional list of words to help the speech recognition algorithm.

Within the `<Gather>` verb, we nested [`<Say>`](/docs/compatibility-api/cxml/reference/voice/say) in order to play some instructions.

When input gathering completes, the script will fetch the URL specified in
`action` and will execute it. If, instead, no input is detected within the
timeout, then the following instructions keep executing: in our case, we say "no input detected" and hang up.

## Reading the User Input

When fetching the Gather's `action` URL, SignalWire includes some parameters such as:

* `From`: the caller's number
* `To`: the callee's number
* `Digits`: DTMF digits gathered from the user, if any
* `SpeechResult`: Speech input gathered from the user, if any
* ...more

With a custom web server, you can read these parameters and, depending on their
value, emit a different cXML document to execute. You can easily test this with
sites such as [https://webhook.site](https://webhook.site), which allow you to view the details for
incoming webhook requests.

## Assigning the Script to a SignalWire Phone Number

The final step is to configure one of your SignalWire phone numbers to answer calls using the cXML Script we just created.
You can do that from the "Phone Numbers" section:

In the Dashboard, open **Phone Numbers** > **Purchased**, select the number, and open its edit page. Under **Inbound Call Settings**, select **+ Assign Resource**, choose the cXML script, and save.

> Use cXML to prompt and receive user input.