Text to Speech
Turn text into natural speech with AI — 20 languages, 17 built-in voices, long-text ready. Free, unlimited.
0 / 10000
Загрузите 3–10 секунд чистой речи, чтобы клонировать этот голос; иначе используется встроенный голос.
- 中文
- Junhao · Chinese male A
- Zhiming · Chinese male B
- Weiguo · Chinese male C
- Xiaoyu · Chinese female A
- Yuewen · Chinese female B
- Lingyu · Chinese female C
- English
- Ava · English female A
- Bella · English female B
- Adam · English male A
- Nathan · English male B
- 日本語
- Soyo · Japanese female A
- Saki · Japanese female B
- Mortis · Japanese female C
- Umiri · Japanese female D
- Mei · Japanese female E
- Anon · Japanese female F
- Arisa · Japanese female G
How AI text-to-speech works
This tool is built on open-source speech generation with an audio-tokenizer + language-model architecture: the system encodes the text into a semantic sequence, generates matching speech codes frame by frame, then decodes them into high-fidelity stereo waveform. Synthesis runs in real time on pure CPU, no GPU required. It auto-detects 20 languages, and long text is split by sentence, synthesized and stitched seamlessly so the whole clip keeps one consistent voice and natural rhythm.
Voice cloning
Upload a 3-10 second sample of clear speech (single speaker, no background music works best) and the model extracts that voice's timbre to read any text you enter in the same voice. Cloning is used only for this task — the reference audio and result are deleted within 24 hours and never used for training or retention. Only clone your own voice or voices you have permission to use.
常见问题
Is this tool free?
Completely free, no registration, no usage limits.
Which languages are supported?
20 languages: Chinese, English, German, Spanish, French, Japanese, Italian, Hungarian, Korean, Russian, Persian, Arabic, Polish, Portuguese, Czech, Danish, Swedish, Greek, Turkish — auto-detected from the text.
Which voices are available?
17 built-in voices: 6 Chinese (3 male / 3 female), 4 English, 7 Japanese. You can also upload a 3-10 second reference audio to clone your own voice.
How long can the text be?
Up to 10,000 characters per task; long text is split by sentence and stitched seamlessly with a consistent voice.
How long does it take?
Usually 10-60 seconds, depending on text length and server load.
Is my data kept?
No. Your input text, reference voice and result are deleted within 24 hours.
Can I use the audio commercially?
Output audio may be used for personal and commercial purposes. Get permission before cloning someone else's voice.