Quick answer: To use ElevenLabs text to speech, sign up for a free account, open the Text to Speech page, choose a voice and a model, paste your script and click Generate. Listen, tweak the voice settings if needed, then download your audio as an MP3. The free plan gives you 10,000 credits a month, which is about 10 minutes of speech.
That is the short version. Below, we walk you through every step, explain what each setting does, show you what the plans include and help you fix the most common problems.
How to use ElevenLabs text to speech in 7 steps
- Create a free account on the official ElevenLabs website (elevenlabs.io).
- Open Text to Speech from the left menu.
- Pick a model. If you are not sure, start with the default.
- Pick a voice from the voice list or the Voice Library.
- Paste your script into the text box.
- Click Generate and listen to the result.
- Download the audio file and use it in your video, podcast or course.
Your first clip takes less than two minutes. Getting it to sound great takes a little more care, and that is what the rest of this guide is about.
Is ElevenLabs free?
Yes, there is a free plan. It is great for testing, but it has limits. Here is what each plan included when we checked the official pricing page in October 2026:
| Plan | Price per month | Credits per month | About this much speech | Commercial use |
|---|---|---|---|---|
| Free | $0 | 10,000 | ~10 minutes | No |
| Starter | $6 | 30,000 | ~30 minutes | Yes |
| Creator | $22 | 121,000 | ~2 hours | Yes |
| Pro | $99 | 600,000 | ~10 hours | Yes |
| Scale | $299 | 1,800,000 | ~30 hours | Yes |
A few things to know:
- 1 credit = about 1 character of text (spaces included) on the standard models. Some fast “Flash” models can cost less per character.
- Credits are shared across all ElevenLabs tools, such as dubbing and sound effects.
- Unused credits roll over for up to two months on paid plans. The free plan does not roll over.
- Commercial use (monetised YouTube videos, client work, ads) needs a paid plan.
Want to know exactly how many credits your project needs? Try our free ElevenLabs credit calculator. Prices change, so always confirm on the official pricing page.
Which ElevenLabs model should you choose?
A model is the “engine” that turns your text into speech. Each one has a different strength:
| Model | Best for | Languages |
|---|---|---|
| Eleven v4 | The most emotional, natural speech. Great for videos, stories and characters. | 90+ |
| Eleven v3 | Dramatic delivery and dialogue between several speakers. | 70+ |
| Multilingual v2 | Calm, steady narration. Very reliable for long audiobooks and courses. | 29 |
| Flash v2.5 | Very fast and cheaper. Best for apps, bots and drafts. | 32 |
Simple rule: use an expressive model (v4 or v3) when emotion matters, Multilingual v2 for long and steady narration, and Flash when speed or cost matters most.
How to pick the right voice
The voice matters as much as the model. Use filters in the Voice Library to narrow the list by language, accent, gender, age and use case (for example “narration” or “conversational”).
- Test with your own script. The default sample sentence hides problems.
- Think about your audience. A calm voice suits a course. An energetic voice suits a short social video.
- Listen for at least 30 seconds. Some voices sound great at first but get tiring.
Not sure where to start? Our AI voice library by use case gives you search terms and starting settings for audiobooks, YouTube, meditation and more.
ElevenLabs voice settings explained
Open the settings panel to change how the voice performs. Change one setting at a time and regenerate the same line so you can hear the difference.
| Setting | What it does | Good starting point |
|---|---|---|
| Stability | Low = more emotion and variety. High = steadier and more even. | 40–60% |
| Similarity | How closely the result sticks to the original voice. | 70–80% |
| Style | Makes the voice’s natural style stronger. Can reduce stability. | 0–20% |
| Speed | Makes speech a little faster or slower. | 1.0 (normal) |
Not every model shows every setting. When you find a mix you like, write it down. Using the same settings every time keeps a series sounding consistent.
How to write a script that sounds human
Most “robotic” results come from the text, not the voice. Before you generate:
- Use short sentences and everyday words.
- Add commas where a person would take a breath.
- Write numbers and symbols the way they are spoken (“fifty percent”, not “50%”).
- Split long text into short paragraphs.
Our free text-to-speech script cleaner turns numbers, prices, dates and acronyms into speakable words in one click. For 12 more tips, read how to make AI voices sound natural.
How to download your audio
Click the download button next to your generation. MP3 works for almost everything: YouTube, podcasts, TikTok and courses. Paid plans unlock higher-quality options, including uncompressed PCM (WAV-style) audio, which is better if you plan to edit or master the sound.
How long will my audio be?
As a rule of thumb, one minute of speech is about 900–1,000 characters, or roughly 150 words. A 10-minute YouTube video needs about 1,500 words. Use our script length calculator to check your script before you spend credits.
Common problems and quick fixes
| Problem | Quick fix |
|---|---|
| Voice sounds flat | Lower stability a little, or try an expressive model. |
| Voice sounds over-acted | Raise stability and lower style. |
| Wrong pronunciation | Spell the word the way it sounds, or use a pronunciation dictionary. |
| Numbers read wrongly | Write them out as words. |
| Rushed delivery | Shorten sentences and add commas or paragraph breaks. |
| Ran out of credits | Generate in small sections and test with short samples first. |
Can you use ElevenLabs audio commercially?
Yes, on a paid plan. The free plan is for personal testing only. If you plan to earn money from a video, publish an audiobook or make content for a client, choose Starter or higher. Community voices in the Voice Library may also have their own terms, so check the voice details before you publish.
Frequently asked questions
Is ElevenLabs text to speech free?
Yes. The free plan gives you 10,000 credits a month (about 10 minutes of speech) for personal use. Commercial use needs a paid plan.
How many languages does ElevenLabs support?
It depends on the model. Eleven v4 supports more than 90 languages, while Multilingual v2 supports 29.
Can I use ElevenLabs on my phone?
Yes. The website works in mobile browsers, and ElevenLabs also offers official mobile apps for some features.
Which ElevenLabs model sounds the most realistic?
The newest expressive models (Eleven v4 and v3) usually sound the most emotional and natural. For long, steady narration, many people prefer Multilingual v2.
Do spaces count as characters?
Yes. Credits are counted on every character you submit, including spaces and punctuation.
Can I clone my own voice?
Yes. Instant voice cloning is available from the Starter plan, and professional voice cloning from the Creator plan. Only clone voices you own or have clear permission to use.
Final thoughts
ElevenLabs is easy to start with: pick a model, pick a voice, paste a clean script and generate. The free plan is perfect for learning, and a paid plan unlocks commercial use and more credits. Start small, test a few voices with your own script, save your favourite settings and you will be making natural-sounding AI voiceovers in no time.
ElevenLabsAIVoice.com is an independent guide and is not affiliated with ElevenLabs. Plans, prices and models change, so always check the official ElevenLabs website for the latest details.
Leave a Reply