Voice cloning for AI voice bots: make the bot speak in your voice
Record 2–3 minutes of speech or upload a file, and in about a minute the voice appears in the bot settings. Recording requirements and a step-by-step guide.
Voice cloning for AI voice bots creates a bot voice from a 2–3 minute recording of one person in the “Your own voice” section. Open it via “Add your own voice” in bot settings, upload or record a sample and click “Create voice”. Once it is “Ready”, pick it in “Voice style” → “My voices”.
Standard voices are shared by all accounts on the service: other companies may call with the same timbre, and a client who hears “yet another robot” hangs up in the first seconds. Voice cloning for AI voice bots creates the bot’s voice from a recording of a real person: your CEO, an expert or your best sales manager speaks, and the bot becomes recognizable. A familiar, lively voice raises the chance that the person listens to the end and gets to a booking or a payment.
Two to three minutes of recording is enough. The voice is ready in about a minute, and creating it costs nothing. Watch the one-and-a-half-minute video, and the step-by-step setup is below.
Why businesses need it
- Online school: the course author’s voice calls webinar sign-ups and reminds them when the live session starts. Participants hear the person they saw on screen.
- Clinic or salon: the administrator’s voice confirms appointments, and at night the bot answers incoming calls in the same voice; that’s how a 24/7 AI receptionist for incoming calls works.
- Real estate: the agent’s voice qualifies leads, and the bot transfers warm clients to that same person through AI voice bot call transfer. The handover sounds natural.
- Branch networks and franchises: one branded voice across all accounts through access for other accounts. You can create up to three voices per account, for example one for sales and one for customer care.
How voice cloning works
- You upload a file or record speech from your microphone right on the page.
- The page checks the recording: whether there is only one voice and no music, noise or sharp sounds. This usually takes 5–15 seconds.
- You click “Create voice”, and the voice takes about a minute to prepare.
- The finished voice appears in “Voice style” in the bot card, under “My voices”. Other sound settings are covered in “How to make an AI voice bot sound human”.
How to create a custom voice for your bot: step by step
- 1Open the sectionBot card → “Settings” → “Add your own voice”. The section opens in a new tab, and the bot card stays where it was.
- 2Name the voiceFill in “Voice name”, for example “Anna - Sales”. Em dashes, commas and quotation marks are not accepted.
- 3Add a recordingClick “Upload a file”: MP3 up to 10 MB or WAV up to 100 MB, from 2:00 to 3:00 long. Or click “Record now” → “Start recording” and read the text from the page.
- 4Wait for the check“Check passed: one voice, clean background.” appears under the player.
- 5Choose accessIn “Who can use the voice”: “Only my account” or “All users”. In the second case, the voice becomes public after moderation (a review by the service).
- 6Confirm and createTick the voice owner’s consent and click “Create voice”. In about a minute, the voice gets the “Ready” tag.
- 7Pick the voice in the botRefresh the bot card page, open “Voice style” → the “My voices” group and click “Save”.
How to record a voice sample for the bot
The recording requirements on the page are simple:
- one speaker for the whole recording;
- lively, natural speech, the way you talk to customers;
- a quiet room and a good microphone or headset;
- no music, echo, street noise or other people’s voices;
- not a phone call recording: its quality is too low.
Record in a room with furniture or curtains: bare walls create echo. Keep the microphone at the same distance, 15–20 cm, and don’t tap on the table. Speak calmly and warmly at a normal pace: monotonous reading produces a monotonous voice. The easiest way is to click “Record now” and read the ready-made sales call text, which is designed for the right length.
If the check finds a problem, the “This recording will not work for a voice” window opens with the reason and a tip.
My voices: access, publishing, deletion
Each voice in the “My voices” list has tags: “Creating”, “Ready” or “Error”, plus “Private”, “Awaiting moderation” or “Public”. Voices belong to the account, so they are visible in all of its projects.
In a voice’s “⋯” menu:
- “Listen to the sample” — the original recording;
- “Access for other accounts…” — share the voice with a branch or partner by account ID, project number or login, up to 50 accounts. Access opens immediately;
- “Make public” or “Make private”;
- “Delete the voice” — in “Replace in scenarios with”, choose the voice the bots will switch to. No bot is left without a voice.
Frequently asked questions
Why did my voice recording fail the check because of several voices?
Why is my own voice missing from the bot’s “Voice style”?
How much does voice cloning cost and how long does it take?
Why is the “Create voice” button greyed out?
Why can’t a partner see the voice shared with them?
What if my bot’s custom voice sounds unnatural?
Read also: How to make an AI voice bot sound human and answer without pauses, AI voice bot for business: how to create a bot and get leads, 24/7 AI receptionist: how to set up a voice bot to answer every call
Everything from this article is already waiting in your account — open the section and get started.
Try LptrackerOur implementation team will set up the system for your business: voice and chat AI bots, funnels, robots, telephony and CRM. Book a meeting at a time that suits you.
Book a meeting