Project AURIONAI Use-Cases Research & Innovation
speech synthesis

OmniVoice

k2-fsa open TTS model; code Apache 2.0, weights CC-BY-NC.

k2-fsak2-fsa
k2-fsa
Levelmodel
Classspeech synthesis
Parameters600M
Contextnot in the notes
ReleasedApr 2026
Opennessopen weights restricted
Runs onconsumer GPU
Statusactive

Lineage

instruct ofcontinued pretrain
Qwen3-0.6B-Base to Qwen3-0.6B: instruct of, high confidenceQwen3-0.6B to OmniVoice: continued pretrain, medium confidenceQwen3-0.6B-BaseQwen3-0.6B-BaseAlibaba (Qwen), Apr 2025Qwen3-0.6BQwen3-0.6BAlibaba (Qwen), Apr 2025OmniVoiceOmniVoicek2-fsa, Apr 2026
Alibaba (Qwen)Qwen3-0.6B is the parent (continued pretrain)medium confidencemodel card
Where this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 3 models. Confidence and evidence are the notes' own.

R&D Topics

Large-scale zero-shot text-to-speechproduced by

Identity

Developer
k2-fsa
Family
not in the notes
Released
Apr 2026 sourcesregistry date
Weights published
–
Record checked
10 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.

Architecture and Size

Architecture
diffusion Diffusion-LM style, base Qwen3-0.6B
Total parameters
600M sources
Active parameters
–
Checkpoint parameters
–
Context length
not in the notes
Input
text
Output
audio
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).

Openness and Access

Openness
open weights restricted
Licence
CC-BY-NC (version not stated in notes) sources
Access routes
open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.

Deployment

Runs on
consumer GPU
Footprint
not in the notes
Runtimes
not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.

Indonesia

Indonesian support
not stated
Regional languages
not in the notes
Jakarta hosting
unknown
Languages
600+ languages claimed; Indonesian not specifically confirmed
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.

Benchmark Results

The notes record no benchmark result for this model.

Industry Use Cases

No industry use case links this model.

Sources

  1. k2-fsa/OmniVoice , Hugging Face huggingface.co https://huggingface.co/k2-fsa/OmniVoice

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "OmniVoice", https://aurion.semai.id/models/k2-fsa-omnivoice/, records of October 2026.