VibeVoice

VibeVoice TTS

Microsoft's multi-speaker long-form model that generates up to 90 minutes with 4 distinct speakers.

Sign up for 5,000 character limit

Wrap your text in SSML tags for precise control:

<speak><prosody rate="slow">Slow speech</prosody></speak>

Tags the selected model understands — click to drop one into your text where it happens:

This model reads plain text, so inline tags are ignored. For tag-based emotion, switch to an expressive model like Orpheus or Bark.

Define custom pronunciations (word = pronunciation):

-12 +12
0.5x 2.0x
Free with Piper, VITS, MeloTTS
Your generated audio will appear here. Choose a model, enter text, and click Generate.
Audio Generated Successfully
0:00
Download Audio Download .srt Link expires in 24h
Free tier: personal use. Commercial license from $5/mo
Love TTS.ai? Tell your friends!

About VibeVoice

VibeVoice from Microsoft is built for long-form, multi-speaker audio. Its 1.5B model can generate up to 90 minutes of speech with as many as 4 simultaneous speakers, using speaker tags to drive multi-turn dialogue — a strong fit for podcasts, audiobooks, and conversations that need speaker consistency across long passages. A separate Realtime 0.5B variant reaches roughly 300ms latency for interactive use. On TTS.ai it covers English and Chinese and accepts up to 50,000 characters per request, so an entire episode can be scripted in one pass.

Best for: Podcasts, dialogues, long-form narration, multi-speaker content

Browse all VibeVoice voices

At a glance

Developer
Microsoft
License
MIT
Tier
standard
Speed
fast
Voice cloning
No
Languages
English, Chinese
Max characters
50000

VibeVoice voices

Speaker 1

English
Standard Neutral

Speaker 1 (Chinese)

Chinese
Standard Neutral

Speaker 2

English
Standard Neutral

Speaker 2 (Chinese)

Chinese
Standard Neutral

Speaker 3

English
Standard Neutral

Speaker 4

English
Standard Neutral

VibeVoice TTS — FAQ

VibeVoice supports up to 4 distinct speakers and up to 90 minutes of continuous output, with speaker tags for multi-turn dialogue — built for podcasts and long-form narration. It accepts up to 50,000 characters per request.

Yes. Alongside the 1.5B long-form model, a Realtime 0.5B variant achieves roughly 300ms latency for interactive use.

VibeVoice is MIT-licensed. It supports English and Chinese and does not currently support voice cloning.
← All voices