ElevenLabs Voice
Let your agent reply with a natural, spoken voice message in chat and as a WhatsApp voice note, using your own ElevenLabs account.
This guide begins with What ElevenLabs Voice does, continues through What it can do for you, How to use it, Connecting your ElevenLabs account, and finishes with Good to know (limits and safety).
What ElevenLabs Voice does
ElevenLabs Voice gives your agent a real voice. It takes the text your agent would normally reply with and turns it into natural, expressive speech, then delivers it as a playable audio message right inside the chat. On WhatsApp, that same audio arrives as a native voice note, exactly like one a person would record.
The skill connects to your own ElevenLabs account (the popular text-to-speech service). You bring your ElevenLabs API key, and the agent uses it to create speech and to see which voices you have. Nothing else in your ElevenLabs account is touched.
In plain terms, it can do two things:
- Read a piece of text aloud and send it back as a voice message.
- List the voices in your ElevenLabs account so you can pick the one that fits your brand.
Because the speech is created in 30+ languages, the same agent can answer an English customer and a Turkish customer, each in a natural-sounding voice.
What it can do for you
Here is what the skill can actually do, in everyday terms:
- Turn any reply into a voice message. Ask the agent to "say" or "read out loud" a greeting, an answer, a reminder or a short announcement, and it sends back a playable audio clip instead of (or alongside) plain text.
- Speak 30+ languages. The default voice model is multilingual, so a message in English, Turkish, Spanish and many other languages all come out sounding natural.
- Deliver audio where your customer is. In web chat the message shows up as an audio player. On WhatsApp it becomes a proper voice note.
- Use any voice from your account. You can list the voices available in your ElevenLabs account and pick a specific one, either as the agent's default or for a single message.
How to use it
- Open the Skill Marketplace in your Botonom dashboard and find ElevenLabs Voice (category: Communication).
- Install it, then assign it to the agent you want to give a voice to.
- Add your ElevenLabs API key (see the next section), and you are ready.
After that, you and your team just ask in plain language. A few realistic examples:
User: Send the customer a short voice note saying their order #4187 shipped today.
Agent: Done. Here is the voice message.
▶ [audio message]
User: Which voices do I have available?
Agent: You have 4 voices on your account: Rachel (calm, female),
Adam (deep, male), Bella (soft, female) and your cloned
"Brand Voice". Want me to set one as the default?
User: Read this welcome out loud in Turkish: "Merhaba, mağazamıza
hoş geldiniz!"
Agent: Here you go.
▶ [audio message]
The agent only creates speech when you clearly ask for audio, a voice note, or for something to be read aloud. Normal text answers stay text.
Connecting your ElevenLabs account
This skill uses your own ElevenLabs API key (this is called BYOK, "bring your own key"). There is no login pop-up to click through; you paste one value.
- Sign in to your ElevenLabs account and copy your API key (it starts with
sk_). - In Botonom, open the ElevenLabs Voice skill settings for your agent and paste it into the ElevenLabs API key field. This value is stored encrypted and is never shown back in full.
- Optionally, set a Default voice (the voice ID from your account) and a Model. If you leave these blank, the skill uses a sensible default multilingual voice.
Good to know (limits and safety)
A few honest boundaries so there are no surprises:
- It only goes one way: text to voice. The skill turns text into speech. It does not transcribe incoming voice notes, and it does not do voice cloning, dubbing or audio editing. Managing or creating voices happens inside ElevenLabs itself; here you simply choose from what you already have.
- Plain text, one message at a time. Each request handles up to about 5,000 characters of plain text (no special markup). Longer content should be split into shorter messages.
- It runs on demand. Speech is created only when someone explicitly asks for audio, never automatically on every reply, which keeps your ElevenLabs usage under control.
- Your data stays in your workspace. The text you send is passed to ElevenLabs to generate the audio, and the finished audio file is hosted on a durable Botonom link so it can play in chat and on WhatsApp. Your API key stays encrypted and is never written into logs or shown to customers.
- Provider stays behind the scenes. Customers just receive a natural voice message; they never see the technical details.