> ## Documentation Index > Fetch the complete documentation index at: /help/llms.txt > Use this file to discover all available pages before exploring further. # Voice notes in your own voice > Clone your own voice once, then send every contact a LinkedIn voice note in your voice that says their name and a line about them. > [!NOTE] > > **Clone your voice turns on with one payment.** Open **Settings > Voice** (or click **Clone your voice** at the top of **Campaigns**) and click **Turn it on for \$20**. That's a one-time payment with **500 voice note credits included** (1 credit = 30 seconds, valid for 3 months from the payment). It turns on for your account as soon as the payment goes through: back from the checkout the page says **Payment received: it turns on for your account in a moment.** and then shows **Clone your voice**. Only owners and admins can pay. A **Message** step can send a LinkedIn voice note in **your own voice** to every contact, with their own name and details. You record your voice once; Kerux makes a copy of it (a "cloned voice"), and each contact's note is spoken from a text you write, with [variables](/help/campaigns/variables.md) like `{{contact.first_name}}` and `{{contact.company}}`. Only **your own** voice, with your consent. Never record or clone someone else. ## 1. Clone your voice (Settings > Voice) **1. Open Settings > Voice** Click **Settings > Voice**. **2. Record yourself** Click **Record** and speak naturally for **1 to 2 minutes** (at least 30 seconds), somewhere quiet. For a text to read, open **Not sure what to say? Read this** and read it aloud. Click **Stop**, listen, then **Use this recording** (or **Record again**). **3. Agree and create** Tick **This is my own voice and I agree to clone it.**, then click **Create my voice**. Kerux keeps who agreed and when. The page then says **Your voice is ready**, with when it was cloned and when you agreed. The recording itself is sent to the voice service once and **Kerux doesn't keep it**. * **One voice per person** in each workspace. To make a new one, click **Delete my voice** first. * **Delete my voice** deletes it in Kerux and at the voice service. Voice notes in your voice then wait until you make a new one. * Recording works in Chrome, Safari and Edge, on a computer or a phone. ## 2. Add a voice note in your voice to a Message step **1. Open the Message step** In your campaign, open a **Message** step. Under **Voice note**, pick **In my voice** (instead of **Recorded**). **2. Write what it says** In **What the voice note says**, write the text, with variables: for example `Hi {{contact.first_name}}, I saw {{contact.company}} is hiring. Worth a quick chat?` About 900 characters fill a minute (the most LinkedIn allows). **3. Listen before you launch** Click **Listen as a contact**: Kerux says the text for the campaign's first contact, in your voice. Previews are free (up to 20 a day per workspace). The step's text box stays optional: with text, the voice note goes with it; without text, only the voice note goes. **Whose voice?** Each note is spoken in the voice of the person who **connected the LinkedIn account that sends it**. If that person hasn't cloned their voice yet, those contacts wait (see below). ## When the campaign runs For each contact, right before the message goes, Kerux: 1. fills in the variables for that contact. If one is empty and has no fallback, the contact **waits** with the usual missing-data message: never a voice note with a blank. Fill in the contact (or set a fallback) and it goes on. 2. speaks the text in the voice, once, and keeps the audio. If the message has to be tried again, the **same** audio is sent: never spoken or charged twice. 3. sends it as a LinkedIn voice note, with the same safety as any message: at most once, checked when LinkedIn's answer is unclear. Each contact's history shows **Voice note made in the sender's voice** and **Voice note in the sender's voice sent**. ## Voice note credits Voice notes use their own credits, apart from AI credits and search credits: * **1 credit = 30 seconds of voice note.** A note up to 30 seconds (about 450 characters) takes **1 credit**; up to a minute (about 900 characters) takes **2**. Credits are taken only when the note is made. * If it can't be made (the voice service fails), the credits **come back** and Kerux tries again later. * Previews (**Listen as a contact**) are free. * **Credits expire 3 months after you buy them.** Credits not used by then are lost and aren't refunded. The oldest credits are used first. * Buy them in **Settings > Billing**, card **Voice notes** (500 credits for US\$20). The card shows your balance, how many credits expire next and when, and every movement, including credits that expired. ## When something waits | What you see | Why | What to do | | - | - | - | | **Waiting for a voice: the person who connected this LinkedIn account hasn't cloned their voice yet** | The sending account's person has no voice in **Settings > Voice**. | They open **Settings > Voice** and clone their voice. The waiting contacts go on by themselves. **Needs attention > Make the voice** opens it. | | **Waiting for voice note credits** | No credits left. Nothing is sent. | **Needs attention > Buy voice note credits** opens **Settings > Billing**. Once you buy, waiting contacts go on at the next pass. | | Missing data | A variable in the voice note's text is empty for that contact. | Fill in the contact, set a fallback, or remove the variable, like any message. | Before you launch, the launch check warns when a sending account's person has **no cloned voice yet**, and when the campaign may need **more voice note credits** than you have. **Fix** opens **Settings > Voice** or **Billing**. ## In the API `GET /v1/voice` (your voice and credits), `GET /v1/voice/credits`, and `POST /v1/workflows/{id}/voice-preview` (Listen as a contact). Cloning or deleting your voice is only done by you, signed in to Kerux (Settings > Voice): `POST /v1/voice` and `DELETE /v1/voice` answer 403 to an API key or an AI tool using one, so nobody can clone or remove a voice in your name from outside. A Message step carries the text as `config.clonedVoice.text`. See the [API reference](/help/developers/api.md). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.