Free AI Text to Speech — 40 Voices in 10 Languages
Paste text, pick a voice, and direct the performance with plain-language instructions. Natural speech in seconds — download MP3/Opus, use commercially.
Generate speech freePick your language, hear it first
Every sample below is real Seed Speech 2.0 output — no studio post-processing. Open a language page for its full voice list and tips.
Direct the voice, not just the words
Most TTS tools stop at picking a voice. Seed Speech 2.0 adds a voice_instruction field — a free-text note to the voice, exactly like directing a voice actor. Write "whisper, as if sharing a secret", "confident keynote speaker", or "on the verge of tears" and the same voice performs your script that way. The instruction is never spoken aloud: it shapes tone, emotion and pacing while your script text stays untouched.
That single field replaces hours of takes: keep the script fixed and iterate only the direction — five tonal variants of one ad read cost a few characters worth of credits.
- 40 preset voices — narrators, hosts, character voices, many bilingual (EN + ZH).
- 10 languages: English, Chinese, Japanese, Spanish, Indonesian, Portuguese, Korean, Italian, German, French.
- Fine controls for speed, pitch and volume — slow down for learning tracks without distortion.
- Export MP3 for universal compatibility or Opus for smaller files.
- Per-character pricing with free signup credits — no subscription needed to start.