Add a byte-level BPE Chinese+English non-streaming zipformer model (#1645)

This commit is contained in:
Fangjun Kuang
2024-12-24 16:56:49 +08:00
committed by GitHub
parent fe3265aa25
commit 08d771337b
5 changed files with 127 additions and 0 deletions

View File

@@ -8,6 +8,27 @@ log() {
echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
}
log "test offline zipformer (byte-level bpe, Chinese+English)"
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-zipformer-zh-en-2023-11-22.tar.bz2
tar xvf sherpa-onnx-zipformer-zh-en-2023-11-22.tar.bz2
rm sherpa-onnx-zipformer-zh-en-2023-11-22.tar.bz2
repo=sherpa-onnx-zipformer-zh-en-2023-11-22
./python-api-examples/offline-decode-files.py \
--tokens=$repo/tokens.txt \
--encoder=$repo/encoder-epoch-34-avg-19.int8.onnx \
--decoder=$repo/decoder-epoch-34-avg-19.onnx \
--joiner=$repo/joiner-epoch-34-avg-19.int8.onnx \
--num-threads=2 \
--decoding-method=greedy_search \
--debug=true \
$repo/test_wavs/0.wav \
$repo/test_wavs/1.wav \
$repo/test_wavs/2.wav
rm -rf sherpa-onnx-zipformer-zh-en-2023-11-22
log "test offline Moonshine"
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-moonshine-tiny-en-int8.tar.bz2

View File

@@ -0,0 +1,61 @@
name: add-new-asr-models
on:
# push:
# branches:
# - new-asr-models
workflow_dispatch:
concurrency:
group: add-new-asr-models-${{ github.ref }}
cancel-in-progress: true
jobs:
add-new-asr-models:
runs-on: ${{ matrix.os }}
name: New asr models
strategy:
fail-fast: false
matrix:
os: [ubuntu-latest]
steps:
- uses: actions/checkout@v4
with:
fetch-depth: 0
- name: Download icefall-asr-zipformer-multi-zh-en-2023-11-22
shell: bash
run: |
d=sherpa-onnx-zipformer-zh-en-2023-11-22
mkdir $d
pushd $d
wget -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/data/lang_bbpe_2000/tokens.txt
wget -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/data/lang_bbpe_2000/bbpe.model
wget -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/exp/decoder-epoch-34-avg-19.onnx
wget -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/exp/encoder-epoch-34-avg-19.int8.onnx
wget -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/exp/encoder-epoch-34-avg-19.onnx
wget -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/exp/joiner-epoch-34-avg-19.int8.onnx
wget -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/exp/joiner-epoch-34-avg-19.onnx
mkdir test_wavs
cd test_wavs
wget -O 0.wav -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/test_wavs/_1634_210_2577_1_1525157964032_3712259_29.wav
wget -O 1.wav -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/test_wavs/_1634_210_2577_1_1525157964032_3712259_55.wav
wget -O 2.wav -q https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/resolve/main/test_wavs/_1634_210_2577_1_1525157964032_3712259_75.wav
popd
tar cvjf $d.tar.bz2 $d
ls -lh $d
rm -rf $d
- name: Release
uses: svenstaro/upload-release-action@v2
with:
file_glob: true
file: ./*.tar.bz2
overwrite: true
repo_name: k2-fsa/sherpa-onnx
repo_token: ${{ secrets.UPLOAD_GH_SHERPA_ONNX_TOKEN }}
tag: asr-models