Describe a conversation — the language, the level, who speaks. You get the script, the speakers and their voices, ready to play. No account needed to try.
Three requests without an account. A free account gives you ten.
3 requests without an account. 10 with one.
MADE WITH KUKARELLA FROM ONE SENTENCE
“Write a phone conversation between a customer whose delivery is three days late and a support agent, about eight exchanges.”
It became ‘A customer calls about a late delivery’ — a phone call in two voices, nine lines.
The assistant wrote this and chose the voices. The accent is your choice: English comes in 14 accents, Spanish in 23. Change a speaker’s voice and every line of that speaker follows.
Trusted by creators from
Save lesson-prep time and captivate your classroom. Turn dense material into engaging audio tutorials and discussions that students actually listen to. Boost accessibility and improve retention, effortlessly
Cut training costs and onboard new hires faster. Create perfectly consistent training modules, IVR prompts, and customer service dialogues in minutes. Scale your communications without scaling your team
Multiply your content, not your workload. Instantly turn scripts and articles into vivid, Podcast-ready audio. Go from a single blog post to a professional-grade audio experience in minutes, not days
Create audio ads that actually convert. Generate compelling product narratives and test multiple ad variations in minutes. Launch global campaigns with multilingual content to drive sales and expand your reach
Whether you teach a language, build a course, or train your own team.
Your course book’s audio was recorded for another class, at a level that fits some of your learners and not the rest. Describe the situation instead: a hotel check-in, a job interview, an angry customer. In about 30 seconds you have a conversation with named speakers, each with a voice, at the level you asked for.
An AI voice that sounds wrong is usually a multilingual voice reading a language that is not its own. Native voices speak one language, in the regional accent you choose: English in 14 accents, Spanish in 23, native voices for Canadian French. There are 2,000+ voices in 152 languages and accents, and every one plays a sample before you decide.
One listening task used to need four windows: the script, a free reader, an audio editor, the worksheet. Here the audio comes out paragraph by paragraph, so each part goes into your slide, your worksheet or your course module on its own. Change one paragraph and only that paragraph is voiced again.
A dialogue at the market is easier to follow with the market in view, and stock photos never quite match the scene. From any project, ask for an image or a short silent clip and it is made from your own text. Or show the assistant a photograph, and it writes the narration from what it sees.
For teams that train their own people
Four things worth knowing before you voice a whole course.
Pick a voice once and every module, role-play and correction is read by the same person. If that voice ever leaves the library — it is rare, and it is another company that makes it — send us recordings you made with it and we clone it back for everyone.
It writes the words well, but it casts from voices that read any language and does not look at what language your text is in. Choose a native voice in the variety your people actually speak, and every line of that speaker follows.
How a voice reads a row of letters or a phone number belongs to that voice, and two voices from the same company can differ. Try yours on one line before you make forty files. Where it still gets it wrong, space the letters out or split the digits.
Audio comes out paragraph by paragraph, so each part drops into the course tool you use, and changing one paragraph re-voices only that paragraph. Colleagues open the same project; one person edits at a time.
One request becomes a conversation with named speakers and their voices. A speaker is switched to a native voice, and every line follows. Then the audio is downloaded paragraph by paragraph for the lesson.
One lesson can have two or three speakers, each with a different voice. Your learners practise on somebody other than you, in the language they are learning.
Type the situation, the language and the level. The script comes back written and cast, and the audio plays a few seconds later. Nothing to record.
Choose a voice once and keep it. Every lesson, every module and every correction is read by the same person, so nothing sounds patched together.
An afternoon’s work, in any language you need. No voice to book, and no new course book to buy.
Try it on something you need this week.
Each one opens the assistant for that app. No account and no card needed.
There are hundreds of these tools now. We have been building this one since 2019, and most of what works in it came from customers telling us what was missing.
“I needed Latin American Spanish, not Spanish from Spain. Everywhere else the words came out mispronounced, or the stress landed in the wrong place. Here I picked the country I needed, and my listeners stopped noticing the voice at all.”
“I teach beginners, and the dialogues I wrote myself were always a little too hard for them. Now I ask for A2 and I get something my class can follow. What used to be my Sunday evening is twenty minutes.”
“A lecture read out loud puts people to sleep, mine included. Two voices talking about the same material holds them to the end. Same words, completely different result.”
“I keep the same two voices across every unit, so the whole course sounds like one course and not like twelve. After three years they are not robots to me any more. They are colleagues.”
“One recording with a voice actor cost me between one and two hundred dollars, and a week of waiting. I have not booked one since. The money matters, but not waiting is what changed how I work.”
“Plenty of these tools can make a voice. What I have not found anywhere else is people who answer, who fix the thing you asked about, and who tell you honestly when something is not possible yet. That is why I am still here.”
“Lots of companies have the technology. Very few work out what the person using it actually needs. This one asks, and then it changes.”
“Our new agents answer the public by phone, so first they hear practice calls in two Quebec French voices, an agent and a caller. When a procedure changes, we update the text and redo the audio ourselves, so no caller gets an old answer.”
“When a policy changes, the night shift has to hear it too, not only the people at the morning meeting. So we write the change once and turn it into audio in an Australian voice. Every shift hears the same words.”
Every voice in the library, commercial rights for the audio you make, credits every month that roll over while you stay, and one or two cloned voices a month. Prime Lite, Prime and Prime Plus differ in how many credits they give, not in which voices they open.
Turning text into audio costs one credit for a character, spaces included. Writing and editing the text is free, and making the same text again with the same voice costs nothing. Asking the assistant to write for you does cost credits, roughly one for every five characters it reads and writes. One minute of a recording costs one credit when it becomes text. An image costs 2 visuals and a short clip 20. A voice you cloned yourself costs three credits a character.
Subscription credits roll over month to month while you stay on a paid plan. Top-up credits never expire. If you move to the free plan, the credits left from your paid plan stay usable for 30 days and then go. Your top-up credits stay.
No. Your projects stay in your account and nothing is deleted. Audio you have already downloaded is yours to keep. Commercial rights end when the paid plan ends.
Take a plan for one month and you have everything it gives for that month. A plan does not stop by itself, so cancel it before the month ends if one month is all you need. A top-up adds credits only; it opens no voices.
In your Profile, yourself, at any time. Your plan keeps working until the end of the time you have already paid for, and you can turn it back on before then. Need an invoice? Take it from Stripe or write to support@kukarella.com.
Ask for your money back within 15 days of your first payment. Write to support@kukarella.com and we process the refund within one business day.
All major credit and debit cards through Stripe, plus Apple Pay and Google Pay where they are supported.
Your plan stops and a 30-day countdown starts. The credits left in your account stay usable in that time. Fix the payment inside the 30 days and your plan comes back with those credits still there.
Receive bonus characters and minutes when you create a Free account on Kukarella!
Receive bonus characters and minutes when you create a Free account on Kukarella!
