ElevenLabs AI Voice Generator on Cabina.AI
Turn any script into natural speech, clone a voice, compose a music track or transcribe a recording with the ElevenLabs AI voice generator. It all happens in one chat: GPT or Claude helps you write the words, and ElevenLabs reads them out right after, with the whole conversation still there.
You can start free on Cabina.AI and skip a separate ElevenLabs subscription. The ElevenLabs online AI voice generator runs on the same plan as every other model, and you can check what each plan covers before you pay a cent.
What You Can Create With ElevenLabs
You get eight audio tools in one place, from spoken scripts to songs and transcripts.
Natural Speech for Long Scripts
With Multilingual v2, ElevenLabs text-to-speech reads long scripts in a natural voice that sounds just as good at the end as it does at the start.
Emotion on Cue
Add audio tags like whispers, laughs or excited to your script, and Eleven v3 acts them out on cue. That way you direct the ElevenLabs TTS voice line by line.
Voice Changer
Record a line yourself and hear it come back in a different voice. Your timing and emotion carry over from the original take.
Voice Design and Voice Library
Describe the age, gender, accent and tone you want, and you'll get three sample voices to pick from. The ElevenLabs AI voice generation platform also has thousands of ready-made voices.
Eleven v4 for Expressive Speech
The newest ElevenLabs model brings more emotion and expression to speech than any version before it. Need it live? Eleven v4 Turbo is fast enough for real-time use.
Flash v2.5 for Speed
Flash v2.5 is the fastest ElevenLabs model. Pick it when you can't keep listeners waiting, for example in live audio.
Eleven Music
Eleven Music turns a short prompt into a song with vocals or an instrumental, anywhere from 3 seconds to 5 minutes long, in MP3 or WAV.
Scribe Transcription
Upload a recording, and Scribe v2 writes it out in any of 90+ languages, puts a timestamp on every word and tells up to 32 speakers apart.
Which ElevenLabs Model to Pick
You can use ElevenLabs online with all six of its voice models in one chat. Each one fits a different job, and you can run the same script through two of them to hear the difference.
| Model | Best for | Languages |
|---|---|---|
| Eleven v4 | Voiceovers with the most emotion and expression | 90+ |
| Eleven v4 Turbo | Real-time speech with about 100 ms response time | 90+ |
| Eleven v3 | Directing emotion with audio tags in the script | 70+ |
| Multilingual v2 | Steady narration across long scripts | 29 |
| Flash v2.5 | The fastest output, at about 75 ms | 32 |
| Voice Changer (Multilingual STS v2) | Turning your recording into another voice | 29 |
How to Clone a Voice With ElevenLabs
Voice Changer is the speech-to-speech model that handles ElevenLabs voice cloning. You upload a sample of a voice, and it builds a clone from that sample. The whole process takes three steps.
ElevenLabs Voice Cloning Tutorial
Open a new chat, choose ElevenLabs and then select Voice Changer from its list of models.
Upload a clean recording of the voice you want to clone. For the best result, ElevenLabs recommends 1 to 2 minutes of clear audio.
Once you send the request, you'll hear speech in the cloned voice that keeps the emotion of the original recording.
Clean audio matters most for accurate ElevenLabs AI voice cloning. Voices with an uncommon accent or a very unusual sound may come out less exact.
ElevenLabs voice cloning features cover two kinds of clones.
If you'd like a second opinion on your clone, try CosyVoice voice cloning, which copies a voice with a different text-to-speech model.
An instant clone needs only a short sample and is ready almost at once.
A professional clone trains its own model on much more audio and sounds nearly identical to the real voice.
How to Start Using ElevenLabs on Cabina.AI
You'll get from sign-up to your first audio file in four steps.
Create a free account. The free plan includes tokens that work across all models, so they double as ElevenLabs free credits for a first test.
Write the script yourself, or ask Gemini or Claude to draft it for you in the same chat.
Switch to an ElevenLabs model. Your conversation history stays put, so the script is right there.
Generate the audio, listen back, then adjust the text and run it again until it sounds right.
Step 3: pick ElevenLabs and a version
Step 4: press play on any voice
Alternatives to ElevenLabs AI
OpenAIAudio 
Whisper 
CosyVoice 
ElevenLabsMusic 
ElevenLabsScribe 
GeminiTts 
SeedAudio 
MiniMaxAudio Use ElevenLabs in Claude, ChatGPT or Cursor Through MCP
If you build with AI, you can call ElevenLabs from the apps you already work in. Cabina.AI connects over 55 AI models, ElevenLabs included, to Claude, ChatGPT, Cursor and Windsurf through its API and Model Context Protocol, or MCP.
One connection gives you text, image, video and audio models, all on one bill instead of separate vendor subscriptions. You type a plain request in your chat or terminal, and ElevenLabs creates the voice without you switching sites. Setup takes one click or a single terminal command, and the MCP setup guide walks you through it.
ElevenLabs Pricing and Free Plan
ElevenLabs subscription pricing comes down to two numbers per version: the token cost of one use and how much each plan covers, from the free plan up.
| Version | Tokens | Free $0.00 | Starter $4.99 | Basic $9.99 | Growth $18.99 | Premium $99.99 |
|---|---|---|---|---|---|---|
| Per generation | ||||||
| Multilingual v2 | 14.47 | 6 | 31 | 69 | 145 | 967 |
| Eleven v3 | 63.23 | 1 | 7 | 15 | 33 | 221 |
| Flash v2.5 | 51.83 | 1 | 8 | 19 | 40 | 270 |
| Voice Changer | 4.87 | 20 | 92 | 205 | 431 | 2,874 |
| Eleven Music | 91.68 | 1 | 4 | 10 | 22 | 152 |
| Per character | ||||||
| Eleven v4 | 0.06 | 1,623 | 7,305 | 16,233 | 34,090 | 227,272 |
| Eleven v4 Turbo | 0.15 | 649 | 2,922 | 6,493 | 13,636 | 90,909 |
| Per second of audio | ||||||
| Scribe v1 | 3.66 | 27 | 123 | 273 | 574 | 3,829 |
| Scribe v2 | 0.02 | 4,166 | 18,750 | 41,666 | 87,500 | 583,333 |
Free$0.00
Starter$4.99
Basic$9.99
Growth$18.99
Premium$99.99
How to read the table
ElevenLabs API pricing bills you by the character. In the table, each version has a fixed token cost per generation, per character or per second of audio, and that cost is the same on every plan.
You don't need to work out ElevenLabs pricing per million characters. Each plan column already tells you how many generations, characters or seconds of audio you get for the money.
For example, the ElevenLabs subscription cost on Starter is $4.99 a month, which covers 31 Multilingual v2 or 92 Voice Changer generations. Any tokens you don't spend on audio work for text, image and video models too.
Pay as You Go is also available, at
for 1800 tokens.
More AI Voice Models to Try
Want to compare? These models are one click away.
OpenAI makes this text-to-speech model, and it's handy for hearing a second set of voices next to ElevenLabs.
Try CosyVoice when you want another multilingual text-to-speech model to test against ElevenLabs.
Whisper from OpenAI goes the other way and turns your recordings into written text.
ElevenLabs FAQ
Is ElevenLabs Free to Try?
Yes. You get 50 tokens on the free plan, which is enough to test a few voices and hear the quality before you spend anything.
How Long Can an ElevenLabs Voiceover Be?
Eleven v4 and Multilingual v2 take up to 10,000 characters per request, and Eleven v3 takes 5,000. Flash v2.5 goes furthest, at 40,000 characters. These limits are per request, so you can split a longer script into several runs.
What Is the Cheapest Way to Pay for ElevenLabs?
The best ElevenLabs pricing comes with an annual plan, which has the lowest cost per token. If you'd rather not commit to a subscription, Pay as You Go is a one-time payment. You're never charged for a request that gets no response.
Do I Need a Separate ElevenLabs Subscription?
No, the plan you already have covers it. If you subscribe, you can cancel anytime and keep full access until the paid period ends. Unused tokens stay on your balance for at least a year.
What Are the Limits of the Free Plan?
ElevenLabs free tier limitations come down to daily and overall caps on some models. Any paid plan removes them, and so does a Pay as You Go payment, for 100 days after you pay.
How Long a Recording Can Voice Changer and Scribe Handle?
Voice Changer works on segments of up to 5 minutes, so split a longer recording into parts. Scribe v2 accepts files up to 3 GB and 10 hours long.
Can I Clone Someone Else's Voice?
ElevenLabs only allows a professional clone of your own voice. To make one, you pass a verification step that proves the voice belongs to you.
Turn Your Next Script Into Speech
Sign up free and hear your first script in an ElevenLabs voice, with GPT, Claude and every other model in the same account.
