Skip to content

Refactor kokoro tts export #2302

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jun 18, 2025
Merged

Refactor kokoro tts export #2302

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
150 changes: 119 additions & 31 deletions .github/workflows/export-kokoro.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@ name: export-kokoro-to-onnx
on:
push:
branches:
- fix-export-kokoro-1.0-2
- refactor-kokoro-2

workflow_dispatch:

Expand Down Expand Up @@ -34,24 +34,94 @@ jobs:
- name: Install Python dependencies
shell: bash
run: |
pip install kokoro "numpy<=1.26.4" onnx==1.16.0 onnxruntime==1.17.1 librosa soundfile piper_phonemize -f https://k2-fsa.github.io/icefall/piper_phonemize.html misaki[en] misaki[zh] torch==2.6.0+cpu -f https://download.pytorch.org/whl/torch
pip install kokoro "numpy<=1.26.4" onnx==1.16.0 onnxruntime==1.17.1 librosa soundfile piper_phonemize -f https://k2-fsa.github.io/icefall/piper_phonemize.html misaki[en] misaki[zh] torch==2.6.0+cpu -f https://download.pytorch.org/whl/torch sherpa-onnx

- name: Run
env:
HF_TOKEN: ${{ secrets.HF_TOKEN }}
shell: bash
run: |
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/espeak-ng-data.tar.bz2
tar xf espeak-ng-data.tar.bz2
rm espeak-ng-data.tar.bz2
cp -a ./espeak-ng-data ./scripts/kokoro/v0.19
cp -a ./espeak-ng-data ./scripts/kokoro/v1.0
cp -a ./espeak-ng-data ./scripts/kokoro/v1.1-zh

git config --global user.email "csukuangfj@gmail.com"
git config --global user.name "Fangjun Kuang"

cd scripts/kokoro
v=${{ matrix.version }}
if [[ $v = "0.19" ]]; then
cd v0.19
./run.sh

if false; then
# generate samples
git clone https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-tts-samples hf
mkdir -p hf/kokoro/v0.19/mp3
./generate_samples.py
pushd hf
git pull
git add .
git commit -m 'add kokoro samples for v0.19'
git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-tts-samples main
popd
rm -rf hf
fi

elif [[ $v == "1.0" ]]; then
cd v1.0
./run.sh

if false; then
git clone https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-tts-samples hf
mkdir -p hf/kokoro/v1.0/mp3

curl -SL -O https://github.com/csukuangfj/cppjieba/releases/download/sherpa-onnx-2024-04-19/dict.tar.bz2
tar xvf dict.tar.bz2
rm dict.tar.bz2

curl -SL -o date-zh.fst https://huggingface.co/csukuangfj/icefall-tts-aishell3-vits-low-2024-04-06/resolve/main/data/date.fst
curl -SL -o number-zh.fst https://huggingface.co/csukuangfj/icefall-tts-aishell3-vits-low-2024-04-06/resolve/main/data/number.fst
curl -SL -o phone-zh.fst https://huggingface.co/csukuangfj/icefall-tts-aishell3-vits-low-2024-04-06/resolve/main/data/phone.fst

./generate_samples.py
pushd hf
git pull
git add .
git commit -m 'add kokoro samples for v1.0'
git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-tts-samples main
popd
rm -rf hf
fi

elif [[ $v == "1.1-zh" ]]; then
cd v1.1-zh
./run.sh

if false; then
git clone https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-tts-samples hf
mkdir -p hf/kokoro/v1.1-zh/mp3

curl -SL -O https://github.com/csukuangfj/cppjieba/releases/download/sherpa-onnx-2024-04-19/dict.tar.bz2
tar xvf dict.tar.bz2
rm dict.tar.bz2

curl -SL -o date-zh.fst https://huggingface.co/csukuangfj/icefall-tts-aishell3-vits-low-2024-04-06/resolve/main/data/date.fst
curl -SL -o number-zh.fst https://huggingface.co/csukuangfj/icefall-tts-aishell3-vits-low-2024-04-06/resolve/main/data/number.fst
curl -SL -o phone-zh.fst https://huggingface.co/csukuangfj/icefall-tts-aishell3-vits-low-2024-04-06/resolve/main/data/phone.fst

./generate_samples.py
pushd hf
git pull
git add .
git commit -m 'add kokoro samples for v1.1-zh'
git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-tts-samples main
popd
rm -rf hf
fi
else
echo "Unknown version $v"
exit 1
Expand All @@ -61,19 +131,39 @@ jobs:
if: matrix.version == '0.19'
shell: bash
run: |
src=scripts/kokoro
src=scripts/kokoro/v0.19

d=kokoro-en-v0_19

mkdir $d
cp -a LICENSE $d/LICENSE
cp -a espeak-ng-data $d/
cp -v $src/kokoro-v0_19.onnx $d/model.onnx
cp -v $src/model.onnx $d/model.onnx
cp -v $src/voices.bin $d/
cp -v $src/tokens.txt $d/
cp -v $src/README-new.md $d/README.md
cp -v $src/../README.md $d/README.md
ls -lh $d/
tar cjfv $d.tar.bz2 $d

ls -lh $d.tar.bz2

- name: Collect results 0.19 (int8)
if: matrix.version == '0.19'
shell: bash
run: |
src=scripts/kokoro/v0.19

d=kokoro-int8-en-v0_19

mkdir $d
cp -a LICENSE $d/LICENSE
cp -a espeak-ng-data $d/
cp -v $src/model.int8.onnx $d/model.int8.onnx
cp -v $src/voices.bin $d/
cp -v $src/tokens.txt $d/
cp -v $src/../README.md $d/README.md
ls -lh $d/
tar cjfv $d.tar.bz2 $d
rm -rf $d

ls -lh $d.tar.bz2

Expand Down Expand Up @@ -219,41 +309,39 @@ jobs:
git config --global user.email "csukuangfj@gmail.com"
git config --global user.name "Fangjun Kuang"

rm -rf huggingface
dirs=(
kokoro-en-v0_19
# kokoro-int8-en-v0_19
)

export GIT_LFS_SKIP_SMUDGE=1
export GIT_CLONE_PROTECTION_ACTIVE=false
for d in ${dirs[@]}; do
rm -rf huggingface

git clone https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/kokoro-en-v0_19 huggingface
cd huggingface
rm -rf ./*
git fetch
git pull
git clone https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/kokoro-en-v0_19 huggingface
cd huggingface
rm -rf ./*

git lfs track "cmn_dict"
git lfs track "ru_dict"
git lfs track "*.wav"
git lfs track "*.onnx"
git lfs track af_dict
git lfs track ar_dict
git lfs track cmn_dict
git lfs track da_dict en_dict fa_dict hu_dict ia_dict it_dict lb_dict phondata ru_dict ta_dict
git lfs track ur_dict yue_dict

cp -a ../espeak-ng-data ./
mkdir -p test_wavs

cp -v ../scripts/kokoro/kokoro-v0_19.onnx ./model.onnx
cp -a ../$d ./

cp -v ../scripts/kokoro/kokoro-v0_19-*.wav ./test_wavs/
git add .

cp -v ../scripts/kokoro/tokens.txt .
cp -v ../scripts/kokoro/voices.bin .
cp -v ../scripts/kokoro/README-new.md ./README.md
cp -v ../LICENSE ./
ls -lh

git lfs track "*.onnx"
git add .
git status

ls -lh

git status

git commit -m "add models"
git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/kokoro-en-v0_19 main || true
git commit -m "add models"
git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/kokoro-en-v0_19 main || true
done

- name: Publish to huggingface 1.0 float32
if: matrix.version == '1.0'
Expand Down
1 change: 1 addition & 0 deletions scripts/kokoro/.gitignore
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,4 @@
espeak-ng-data
voices.json
voices.bin
README-new.md
Expand Down
10 changes: 3 additions & 7 deletions scripts/kokoro/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,10 +1,6 @@
# Introduction

This folder contains scripts for adding meta data to models
from https://github.com/thewh1teagle/kokoro-onnx/releases/tag/model-files

See also
https://huggingface.co/hexgrad/Kokoro-82M/tree/main
Please see also
https://huggingface.co/hexgrad/Kokoro-82M
and
https://huggingface.co/spaces/hexgrad/Kokoro-TTS

https://huggingface.co/hexgrad/Kokoro-82M/discussions/14
117 changes: 0 additions & 117 deletions scripts/kokoro/add_meta_data.py
View file
Open in desktop

This file was deleted.

Loading
Loading