> ## Documentation Index > Fetch the complete documentation index at: https://docs.callers.ai/llms.txt > Use this file to discover all available pages before exploring further. # Voices and sounds > Pick a voice for your agent, tune how it sounds, clone your own voice, and add background sounds and noise cancelation to calls. The voice is the first thing a contact notices about your agent. Callers gives you a library of ready-made voices from several providers, lets you clone a voice from your own recordings, and adds optional background sounds and noise cancelation to make calls sound natural. You manage voices in two places: | Where | What you do there | | - | - | | **Voices & Sounds** (sidebar, under Resources) | Browse and preview the voice library, create or clone voices, edit their settings, and manage background sounds. | | Agent editor → **Agent** (header button) | Choose the language, dialect, and voice for one agent, and set background sound, noise cancelation, and other voice behavior for that agent. | > [!NOTE] > > The **Voices & Sounds** page is available to account admins. ## Browse the voice library Open **Voices & Sounds** from the sidebar. The page has two tabs: **Voices** and **Background Sounds**. On the **Voices** tab you can: * **Search** voices by name. * **Preview** a voice by clicking its card. Edit the sample text box to hear the voice say your own words. The sample text follows the language you filter by; use the restore option to return to the default text. * **Filter by language** to see only voices that support a given language or dialect. * Open **Filters** to narrow the list: | Filter | Options | | - | - | | **Voices** | **Callers** (built-in voices), **My Voices** (voices your account created), **Any** | | **Gender** | **Male**, **Female**, **Any** | | **Emotions support** | **Yes**, **No**, **Any**. Emotion-ready voices can use [emotion tags](https://docs.callers.ai/documentation/voice/conversation-dynamics.md#emotions-and-emotion-tags). | Built-in voices are marked as platform voices and can't be edited or deleted. Voices your account created have a menu with **Edit** and **Delete**. ![Voices and Sounds page on the Voices tab with a grid of voice cards showing each voice's name, languages, and Emotions support, plus Search, Sample text, Language, Filters, and Create Voice controls](https://mintcdn.com/voxiaai/wqnSuR-gwKpmU86L/images/documentation/voice/voices-library.png?fit=max\&auto=format\&n=wqnSuR-gwKpmU86L\&q=85\&s=99fbe4660514000eeb3e9138b317a395) The Voices tab of Voices & Sounds, with search, sample text, the Language filter, Filters, and Create Voice. ## Choose a voice for an agent **1. Open the agent's voice settings** Open the agent and click **Agent** in the editor header (see [Agent settings](https://docs.callers.ai/documentation/build-your-agent/agent-setup.md#agent-voice-and-conversation)). **2. Pick a language and dialect** Select a **Language**. For languages with several dialects, choose the dialect (for example, English - United Kingdom). The dialect controls both speech recognition and which voices are offered. See [Supported languages](https://docs.callers.ai/documentation/using-callers/supported-languages.md). **3. Pick a voice** Open **Voice**. The list shows only voices that support the selected dialect, grouped by gender. Voices marked **Emotions** support emotion tags. **4. Preview it** Type a line from your script in **Sample text** and click the speaker icon to hear it in the selected voice and dialect. **5. Save** Save the agent. Make a [test call](https://docs.callers.ai/documentation/build-your-agent/testing-your-agent.md) to hear the voice on a real phone line. > [!TIP] > > If the voice you want doesn't appear in the list, it doesn't support the selected dialect. Change the dialect, choose a multilingual voice, or create your own voice with **Support all languages** turned on. Text agents (SMS, WhatsApp, email) only have a **Language** setting. Voice options apply to voice agents. ## Voice settings Each voice you create has settings that shape how it speaks. You set them when you create the voice and can change them later with **Edit** on the voice card. | Setting | What it does | Range | Default | | - | - | - | - | | **Speed** | Speaking rate, from **Slower** to **Faster**. | −1 to 1 | 0 | | **Stability** | Lower values sound more expressive and variable; higher values sound more consistent. ElevenLabs voices only. | 0 to 1 | 0.5 | | **Similarity** | How closely the output matches the original voice samples. ElevenLabs voices only. | 0 to 1 | 0.75 | | **Support all languages** | Offer the voice for every language and dialect. Turn it off to pick specific **Languages**. | On / Off | On | | **Also create for sub-accounts** | Copy the new voice to your sub-accounts. Shown only in parent accounts. | On / Off | Off | > [!TIP] > > Start with the defaults and change one setting at a time. A small speed change (for example 0.1) is usually enough. Very low stability can make long calls sound inconsistent. ## Clone a voice Cloning creates a voice from recordings of a real speaker, so your agent can sound like your brand voice or a member of your team. > [!WARNING] > > Only clone a voice you have the right to use. Get clear, written consent from the speaker before you upload their recordings. Some providers require a recorded consent statement as part of the upload. **1. Start a new voice** Go to **Voices & Sounds** → **Voices** and click **Create Voice**. **2. Name the voice and set its gender** Enter a **Voice name** (at least 5 characters) and choose male or female. **3. Choose a provider** Select the **Provider**: ElevenLabs, Cartesia, or Google Cloud. The provider decides which files you need (see the table below) and which languages the voice can use. **4. Upload your recordings** Add the audio files on the left side of the dialog. You can play each file back before saving. **5. Set languages and voice settings** Keep **Support all languages** on, or turn it off and choose the languages this voice should be offered for. Adjust **Speed** (and **Stability** and **Similarity** for ElevenLabs). **6. Save** Click **Save Voice**. The new voice appears under **My Voices** and in the agent's **Voice** list for the languages it supports. ![Create voice dialog with a drop zone for 1 to 10 MP3 or WAV files on the left and, on the right, Voice name, Male and Female options, Provider set to ElevenLabs, a Support all languages toggle, Speed, Stability, and Similarity sliders, an Also create for sub-accounts toggle, and a Save Voice button](https://mintcdn.com/voxiaai/wqnSuR-gwKpmU86L/images/documentation/voice/create-voice-modal.png?fit=max\&auto=format\&n=wqnSuR-gwKpmU86L\&q=85\&s=8651b90af756953d862b6989d476a241) The Create voice dialog: drop your samples on the left, set the name, provider, and voice settings on the right. ### File requirements by provider | Provider | Number of files | Formats | Max size per file | Notes | | - | - | - | - | - | | **ElevenLabs** | 1–10 | MP3, WAV | 11 MB | More varied, clean samples give a better clone. | | **Cartesia** | Exactly 1 | FLAC, MP3, MPEG, MPGA, OGA, OGG, WAV, WEBM | 16 MB | One clear sample of the speaker. | | **Google Cloud** | Exactly 2 | WAV, PCM, MP3, M4A | 10 MB | One file name must contain `reference` (the voice sample) and the other `consent` (the speaker's recorded consent). | ### Recording tips * Record in a quiet room with no music, echo, or other speakers. * Use a consistent microphone distance and speaking style that matches how the agent should sound on calls. * Speak naturally, in the language the agent will use. * Trim long silences at the start and end of each file. ### Add a voice by provider ID If a voice already exists in your provider account, Callers can add it by its provider voice ID with no audio upload. Contact [support@callers.ai](mailto:support@callers.ai) to add a voice this way. ### Manage voices with the API You can list, create, clone, update, and delete voices with the API (`GET /voices`, `POST /voices`, `POST /voices/voice-clone`, `PUT /voices/{id}`, `DELETE /voices/{id}`). See [Voices and cloning](https://docs.callers.ai/api-reference/guides/voices-and-cloning.md). ## Background sounds A background sound plays quietly under the agent's voice, for example office chatter, so calls feel like they come from a real workplace. > [!NOTE] > > Background sounds play only on calls made through a [SIP integration](https://docs.callers.ai/documentation/integrations/sip.md). The setting appears in the agent only when SIP is enabled for your account. ### Add a background sound to an agent **1. Open the agent's voice settings** Open the agent and click **Agent** in the editor header. **2. Pick a sound** Expand **Background Soundscape** and choose a sound, or **None** to turn it off. **3. Set the volume** Drag **Background volume level** between **Quiet**, **Moderate**, and **Loud**. The default is 50%. ### Upload your own sound Go to **Voices & Sounds** → **Background Sounds** and click **Upload sound**. ![Background Sounds tab showing a search field, a Sound filter, an Upload sound button, and cards for built-in sounds such as Café Vibes, Call Center Buzz, and Busy Office](https://mintcdn.com/voxiaai/wqnSuR-gwKpmU86L/images/documentation/voice/background-sounds-tab.png?fit=max\&auto=format\&n=wqnSuR-gwKpmU86L\&q=85\&s=38987193650eeb5d0f21395994c9e2d7) The Background Sounds tab with the built-in sounds and the Upload sound button. | Requirement | Value | | - | - | | Format | MP3 | | Max file size | 10 MB | | Length | 30 seconds to 5 minutes | | Name | Up to 100 characters | Use the **Sound** filter to show **Platform** (built-in) sounds, **Uploaded** sounds, or **Any**. Uploaded sounds can be renamed or deleted. Deleting a sound removes it from every agent that uses it. > [!TIP] > > Pick a sound that loops cleanly and has no speech in a language your contacts understand, so it doesn't compete with the agent. Keep the volume low; contacts should barely notice it. ## Noise cancelation **Noise Cancelation** reduces unwanted sounds on the contact's side, such as chatter and keystrokes, so the agent hears the contact more clearly and is interrupted less by background noise. Set it in the agent under **Agent** → **Noise Cancelation**. The slider runs from **Basic** to **Extreme**; the default is off (the far left of the slider). | When to use it | Suggested level | | - | - | | Contacts call from quiet places | Off or **Basic** | | Contacts often call from cars, streets, or busy offices | **Moderate** to **Standard** | | The agent stops talking often because of background noise | Increase one step at a time and test | > [!WARNING] > > Strong noise cancelation can also cut quiet or distant speech. Raise the level gradually and check transcripts after each change. > [!NOTE] > > If you don't see **Noise Cancelation** in your agent, contact [support@callers.ai](mailto:support@callers.ai). ## Troubleshooting > **My cloned voice doesn't appear in the agent's voice list** > > The voice list only shows voices that support the agent's selected dialect. Edit the voice and turn on **Support all languages**, or add the dialect under **Languages**. Also check that the voice's provider supports that dialect. > **The clone upload is rejected** > > Check the file count, format, and size for the provider in the table above. For Google Cloud, upload exactly two files: one with `reference` in its name and one with `consent`. > **The agent's voice changed after I switched language** > > Each dialect offers its own set of voices. If the current voice doesn't support the new dialect, Callers selects the first available voice. Pick the voice again after changing the language. > **I can't hear the background sound on test calls** > > Background sounds play only on calls that go through a SIP integration. Test with a number connected over SIP. ## Related > **[Conversation dynamics](https://docs.callers.ai/documentation/voice/conversation-dynamics.md)** > > Patience, interruptions, and emotions. > **[Supported languages](https://docs.callers.ai/documentation/using-callers/supported-languages.md)** > > Languages, dialects, and multilingual mode. > **[Pronunciation dictionary](https://docs.callers.ai/documentation/voice/pronunciation-dictionary.md)** > > Control how the agent says names and terms. > **[Voices and cloning API](https://docs.callers.ai/api-reference/guides/voices-and-cloning.md)** > > Manage voices programmatically. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.