Downloads · 30 days
0
ShiWarai/kokoro-rknn-ru
kokoro-rknn-ru is a text-to-speech model from ShiWarai. Use it when you need text read aloud. The card lists the license as mit.
Pre-built inference packs for kokoro-rknn — Kokoro-82M Russian fine-tune (zaakirio/kokoro-ru) with NPU-accelerated decoder (.rknn).
Downloads · 30 days
0
Access
Public
Updated Aug 9, 2026
Repo size
486 MB
Likes
0
Public
Click a slice to open those files.
.rknn260 MB · 54%
From the Hugging Face model README
Pre-built inference packs for kokoro-rknn — Kokoro-82M Russian fine-tune (zaakirio/kokoro-ru) with NPU-accelerated decoder (.rknn).
manifest.json
packs/
base/ — sveta, masha (female)
dima/ — dima (male)
Each pack contains: kokoro_encoder.onnx, har_generator.onnx, kokoro_decoder.rknn, config.json, voices_npy/*.npy.
Download a pack into your models/ directory, then run kokoro-cli or kokoro-server from the kokoro-rknn repo (see manifest paths).
huggingface-cli download ShiWarai/kokoro-rknn-ru packs/base --local-dir ./models/base
Or clone with Git LFS:
git lfs install
git clone https://huggingface.co/ShiWarai/kokoro-rknn-ru
| Pack | Voices | Gender |
|---|---|---|
base | sveta, masha | female |
dima | dima | male |
Note: voices from different packs require different decoder weights — switch the full pack, not just the .npy file.
librknnrt)