All models
realtime

CosyVoice 2

Apache-2.0

Alibaba's real-time TTS, 150ms streaming latency

CosyVoice 2 from Alibaba's FunAudioLLM team is engineered for real-time use: streaming synthesis at roughly 150ms first-packet latency with quality nearly identical to offline mode. It supports zero-shot cloning, cross-lingual synthesis, dialect control, and natural-language instructions over emotion and style. CosyVoice 3, a newer 0.5B Apache-2.0 model in the same family, has since been released.

Looking for the newer release? See CosyVoice 3 → These recordings remain labeled with their original version.

Hear CosyVoice 2

CosyVoice 2 · Default · Same three scripts across the directory.

Run it yourself

Best for: Alibaba's real-time TTS, 150ms streaming latency

Permissive model license: Version-specific model card and repository agree on Apache-2.0. Review the exact tokenizer and reference recordings used in a deployment.

Code: Apache-2.0. Weights: Apache-2.0 (CosyVoice2-0.5B). License sources checked 2026-09-20. Source 1Source 2Source 3

Before you choose: Compare the samples with your own use case. Check the linked documentation for code, weight, and voice license terms.

Details awaiting a fresh review. Hardware figures are estimates unless a benchmark is linked.

Compare with similar

Compare all →
Orpheus TTSListen side-by-side →XTTS v2Listen side-by-side →Fish Speech 1.5Listen side-by-side →

Choose an alternative to hear both models together.

Voices

1 voice
Neutral
Defaultdefault
—

The scripts

Neutral
“The quick brown fox jumps over the lazy dog near the riverbank.”
Emotional
“I can't believe you actually did it. This is incredible!”
Numbers & Dates
“On March 14th, 2025, the team raised $4.2 million at a 38% margin.”

Run CosyVoice 2

pip install cosyvoice

Sources & setup details

CosyVoice 2 FAQ

Do I need a GPU to run CosyVoice 2?

The directory lists roughly 6 GB of VRAM as a historical estimate. Check the model documentation for your checkpoint and runtime.

Can CosyVoice 2 clone voices?

Yes — CosyVoice 2 supports voice cloning from reference audio.

What license is CosyVoice 2 released under?

CosyVoice 2 is released under the Apache-2.0 license. Always check the repository for the exact terms — some models license code and weights separately.

What languages does CosyVoice 2 support?

CosyVoice 2 supports Chinese, English, Japanese, Korean, Cantonese, WUU, SICHUANESE.

Is there a hosted CosyVoice 2 API?

OpenSpeech Cloud is accepting production-access interest for CosyVoice 2, but this model is not currently hosted in Cloud. Use the official repository for current deployment options.