k2-fsa / text-to-speech updated 2 months ago

OmniVoice

Name: OmniVoice
Author: k2-fsa

OmniVoice is a massively multilingual zero-shot text-to-speech (TTS) model supporting over 600 languages. Built on a novel diffusion language model-style architecture, it delivers high-quality speech with superior inference speed, supporting voice cloning and voice design.

Params

610 M

Context

—

Downloads 30d

1.3 M

Likes

1,408

Licence: unknown Not gated SAFETENSORS 646 languages View on Hugging Face ↗

Download history

daily snapshots · 55 days

▲ 380 K in the last 30 days (40.7%)

1.3 M765 K

Jul 28Aug 15Sep 2Sep 20

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

File	Quant	Size	Est. VRAM	Verdict on RTX 4090 · 24 GB
model.safetensors	f32	3.3 GB	4.2 GB	✅ Runs comfortably

Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Specifications

Architecture: OmniVoice
Parameters: 610 M
Tensor type: F32
First seen on the Hub: 2026-03-30
Base model: Qwen3-0.6B
Training datasets: undisclosed
Added to our catalog: 2026-07-28

Family

Base model and the most-downloaded derivatives in the catalog.

Base Qwen3-0.6B Qwen 23.0 M

Compare with any text-to-speech model

Popular comparisons

Kokoro-82M 11.6 M XTTS-v2 7.3 M audio.cpp-gguf 4.3 M Qwen3-TTS-12Hz-1.7B-CustomVoice 2.6 M chatterbox 1.8 M F5-TTS 949 K

Appears in

Best open-source TTS