Import your own voice
Full guide to make your agent speak with your brand voice: an ElevenLabs account with professional cloning, the cloning itself, the share link and the import into Enlassa AI.
A custom voice makes your agent speak with your brand voice: yours, a voice actor, or someone on your team. The voice is created in ElevenLabs (the voice provider) and brought into Enlassa AI with a share link. This guide takes you from zero to your agent speaking with that voice.
Step 1 — Create your ElevenLabs account with professional cloning
- Sign up at elevenlabs.io
Use your business email, not a personal one: the account becomes the owner of the voice.
- Subscribe to the Creator plan or higher
It is the first plan that includes professional cloning. The free and Starter plans only include instant cloning, which does not work for this. Current prices are at elevenlabs.io/pricing.
Step 2 — Prepare the audio
- Record at least 30 minutes; about 3 hours is ideal
That is what ElevenLabs recommends: the more audio, the closer the clone.
- One person and one style
Record in the tone the agent will use: a friendly phone-support tone, not a radio ad. Mixing styles makes the result worse.
- Clean audio
No music, background noise, echo or other voices. A decent microphone in a quiet room is enough.
- Speak the way your calls will sound
Read texts similar to what the agent will say (greetings, questions, confirmations) with the same accent as your customers.
Step 3 — Create the professional clone in ElevenLabs
- Create a new voice of type “Professional Voice Clone”
From the voices section of ElevenLabs (menu names may change a little over time).
- Upload the audio from step 2
Give the voice a name you will recognize.
- Verify the voice
ElevenLabs asks you to read a short text aloud within a time limit to confirm the voice belongs to whoever clones it. It must be done by the same person in the recordings. If verification does not match, ElevenLabs support resolves it.
- Wait for training
ElevenLabs estimates about 3 to 6 hours, sometimes longer. It notifies you when the voice is ready.
Step 4 — Generate the share link
- Open “My Voices” in ElevenLabs
Find the voice you just created.
- Click the three dots (More actions) and choose “Share Voice”
- Turn on “Anyone with the link”
ElevenLabs generates the share link right then.
- Copy the full link
It must contain /share/ (something like elevenlabs.io/…/share/…). The address of the voice page does not work: it has to be the share link.
Step 5 — Import it into Enlassa AI
- Open your agent and go to the voice
In the voice picker, click Import voice.
- Paste the link into “Voice link (share link)”
- Give it a name
The one you will see in the picker, for example “Maria · Warm”.
- Click “Import voice”
The voice stays private to your organization: nobody else sees it. It shows up in the voice picker of all your agents.
- Select it in the agent and save
Step 6 — Test it on a call
Make a test call and listen above all to the questions: some voices inflect them well and others say them flat. If your agent asks important questions (confirming a payment, a date), that test is worth more than any tweak.
How many you can have and what they consume
- The number depends on your plan. If you reached the limit, the platform tells you when importing: hide a custom voice you no longer use or upgrade your plan.
- A custom voice counts as a premium voice: minutes of calls with that voice take a little more from your plan. You can see it in Billing.
- Hiding a voice frees its slot. If you import it again later, it takes a slot again.
It says the link is not valid.
Check that it is the share link (it contains /share/) and not the address of the voice page, and that “Anyone with the link” is turned on in ElevenLabs.
It says my plan does not include custom voices.
Your current plan does not include custom voices or you already used them all. You can see what each plan includes in Pricing and change plans from Billing.
Can I use an instant voice clone?
No. ElevenLabs does not allow sharing those voices outside your account, so there is no link to paste. You need to create it as a professional clone.
Who can import a voice?
An administrator of your organization, or a custom role with permission to manage voices.