> ## Documentation Index > Fetch the complete documentation index at: https://docs.cognigy.com/llms.txt > Use this file to discover all available pages before exploring further. # Voice > Learn how voice settings define how an agent speaks and listens on a phone call. > [!NOTE] > > This feature is in beta. We encourage you to try it out and provide us with feedback. *Voice* settings define how an agent speaks and listens on a phone call. You set them for each agent. A consistent voice and predictable call behavior make conversations sound natural. Voice settings are the defaults that every call starts with. Each new call starts from these defaults, even if a voice tool changed some of them during the previous call. If you leave a setting empty, the call uses the value from its Voice application. A value that you set overrides the application value. If you turn on [speech to speech](https://docs.cognigy.com/ai/agents/develop/voice/speech-to-speech.md) for an agent, the agent doesn't show these voice settings. ## Prerequisites Set up a speech provider and its connection in your project. * Customers with NiCE-integrated environments have preconfigured speech providers. * All other customers must configure speech providers in **AI & Voice Providers**. For more information, see [Speech and telephony](https://docs.cognigy.com/ai/administer/access/speech-and-telephony.md). ## Working with voice To configure voice settings for an agent, follow these steps: 1. In the navigation menu, go to **Design** > **Agent**. Select the agent. 2. On the **Configuration** tab, go to the **Voice** section. 3. Select a **Text-to-speech provider**. This is the service that the agent speaks with. 4. Select a **Speech-to-text provider**. This is the service that the agent listens with. 5. Select a **Text-to-speech language**. To find a language, enter its name or code in the search field. 6. Select a **Voice**. The **Voice** field appears after you select a language. To narrow the list, search by name or gender, or use the **Filter voices** options to show all, male, or female voices. 7. To hear the voice, click **Play voice sample**. Save changes. ## Voice studio Voice studio holds all voice settings of an agent in one place: the voice, the speech recognizer, the greeting, turn taking, and listening. To open it, click **Voice studio** in the **Voice** section of the agent's **Configuration** tab. It opens in a full-page view. > [accordion] Voice > > | Setting | Description | > | - | - | > | **Text-to-speech provider** | The text-to-speech provider. The provider that your project uses by default shows a **Default** tag. A provider that NiCE includes shows an **Included** tag. | > | **Text-to-speech language** | The language of the voice. The agent answers in this language unless the caller clearly speaks another one. | > | **Voice** | The voice that the agent speaks with. | > [accordion] Recognizer > > | Setting | Description | > | - | - | > | **Speech-to-text provider** | The speech-to-text provider. | > | **Speech-to-text language** | The language that the agent listens for. If you select a speech-to-text provider and leave this field empty, Cognigy uses the text-to-speech language. | > [accordion] Greeting > > Defines what the agent says when it picks up the call. Select one of the following options: > > * **Agent decides**: the agent generates the greeting. > * **User defined**: the agent says the text that you enter in **Greeting text**. To hear it, click **Play greeting**. You can't save the settings while **Greeting text** is empty. > > If the agent was set to stay quiet and wait for the caller, neither option is selected. > [accordion] Turn taking > > | Setting | Description | Default | > | - | - | - | > | **Interruptions** | Defines when the caller can interrupt the agent:If the agent uses another number of words, the field shows **Custom** and the number of words. Cognigy leaves this value unchanged. | Natural | > | **Check in after** | The number of seconds of silence before the agent checks in. Select 3 to 20 seconds. | 8 | > | **Check in up to** | The number of check-ins before the agent ends the call. Select 0 to 5. With 0, the agent doesn't check in. | 2 | > [accordion] Listening > > | Setting | Description | Default | > | - | - | - | > | **Callers can use the keypad** | Listens for dual-tone multi-frequency (DTMF) keypad digits from the caller. | Off | ## Voice tools Voice tools let an agent act during a call. To add a voice tool, go to the agent's **Tools** tab and click **Add tool**. In the **Add to this agent** dialog, select the **Voice tools** tab. For more information, see [Tools](https://docs.cognigy.com/ai/agents/develop/tools.md). In Cognigy, you can use the following voice tools: | Tool | Description | | - | - | | **Hang up** | The agent says a closing sentence and ends the call. | | **Transfer call** | The agent transfers the call to a Session Initiation Protocol (SIP) address that you set. The agent decides when to transfer, but not where. | | **Play audio** | The agent plays one of the recordings that you configure, from a URL that you host. | ## How voice studio and voice tools work together Voice studio sets the defaults for each call, including the voice, greeting, how the agent handles interruptions and silence, and keypad input. Voice tools let the agent take actions during a call, such as ending it, transferring it, or playing audio. Some voice tools can also change call settings while the call is in progress. These changes don't affect future calls. Each new call starts with the defaults set in voice studio. ## More information * [Brand persona](https://docs.cognigy.com/ai/agents/develop/brand-persona.md) * [Tools](https://docs.cognigy.com/ai/agents/develop/tools.md) * [Getting started with voice flow agents](https://docs.cognigy.com/ai/agents/develop/flow-agents/voice.md)