Text to speech with AI, free
No signup · your text stays local · no character limit
- 100% free
- No signup
- Text stays local
- No character limit
Turn any text into audio with realistic neural voices, 100% free. Choose from 28 American and British English voices, adjust the speed and download the WAV. The Kokoro-82M model (82 million parameters, Apache 2.0 license) runs directly in your browser with WebGPU or WASM: your text never leaves your device, there is no character limit and no signup required.
How it works in 3 steps
Type or paste your text
Write directly or paste the text you want to convert to audio. There is no character limit.
Pick a voice and speed
Choose from 28 English voices (American and British), male and female, and set the speed from 0.5× to 2×.
Generate and download
Hit Generate. The model downloads once and stays cached. When it's done, play the audio or download it as WAV.
Why this tool
100% free
No payment, no credit card, no trial that charges you later.
No signup
No email, no password, no spam. Just open it and use it.
Text never leaves your device
The AI model runs in your browser. Your text is never sent to any server.
No character limit
Paste a full article, a book chapter or an entire script. No daily quota, no per-session limit.
28 neural voices
American and British English voices, male and female, with natural quality.
Speed control
Adjust from 0.5× (slow) to 2× (fast) for audiobooks, studying or production.
Who it's for
Content creators
Generate voiceovers for YouTube, TikTok, Instagram Reels or podcast intros without hiring a narrator.
Students
Turn notes, summaries or articles into audio to review while walking or exercising.
Writers and editors
Listen to your text out loud to catch rhythm issues, repetitions or awkward phrasing.
People with visual impairments
Convert any text to audio to listen comfortably, without relying on a screen reader.
Teachers
Create audio material for your classes: readings, instructions or listening exercises.
Developers and makers
Generate audio for prototypes, demos, voice assistants or bots without paying for an API.
Comparison with other text-to-speech tools
Most text-to-speech tools have very limited free tiers or require an account. Here is the honest comparison:
- ElevenLabs: free up to 10,000 characters/month with 3 voices. Then from $5/month for 30,000 characters. Here there is no limit.
- Murf AI: free with 10 minutes of generation and a watermark on the audio. Plans from $26/month. Here there is no watermark and no limit.
- NaturalReader: basic voices free with limited use; premium voices from $10/month. Here all voices are free.
- TTSMaker: free with a 20,000 character/week per-session limit. Here there is no weekly or per-session quota.
Frequently asked questions
Is it really free or is there a hidden limit?
It is 100% free, with no character limit, no daily quota and no pro version. The AI model runs in your browser, so it costs us nothing per use. No catch.
Is my text sent to a server?
No. The Kokoro-82M model downloads once to your browser and everything is processed locally. Your text never leaves your device.
What voices are available?
28 English voices: 14 American and 14 British, both male and female. The default voice (Heart) has the best overall quality. Currently only English is supported; Spanish and other languages will come in a future update.
Can I generate audio in Spanish?
Not yet. The current model version (Kokoro-82M v1.0) only includes English voices. We are working on adding Spanish. If you paste Spanish text, the model will try to read it with English pronunciation and it won't sound right.
How much text can I convert at once?
There is no limit. You can paste a full article, a book chapter or a script. The text is split into chunks internally and processed one by one.
How long does it take to generate the audio?
It depends on your device. With a compatible GPU (WebGPU), a paragraph generates in seconds. Without a GPU (WASM mode), it takes longer but still works. The first time the model needs to download (~200 MB with GPU, ~90 MB without), then it stays cached in your browser.
What is the difference between WebGPU and WASM?
WebGPU uses your device's graphics card and is much faster. WASM works in any modern browser but uses only the CPU, so it is slower. The tool detects which one to use automatically.
What format is the audio downloaded in?
WAV (PCM 16-bit, 24 kHz, mono). It is an uncompressed format compatible with any audio editor or player. If you need MP3, you can convert it with any conversion tool.
How much does ElevenLabs charge for text to speech?
ElevenLabs offers 10,000 free characters per month (about 2-3 pages). After that, the cheapest plan is $5/month for 30,000 characters. This tool has no character limit and is free.
Does it work on mobile?
Yes, in any modern browser. On phones the processing is slower because they lack a dedicated GPU, but it works. Chrome on Android usually gives the best performance.
Need the opposite? Convert audio to text. Go to Audio to Text