8ee02c28b0
* code refactoring * Remove reference files * Update README and CI * small fixes * fix style issues * add style check for CI * fix style issues * remove kaldi-native-io
Introduction
See https://github.com/k2-fsa/sherpa
This repo uses onnxruntime and does not depend on libtorch.
We provide exported models in onnx format and they can be downloaded using the following links:
- English: https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13
- Chinese:
TODO
NOTE: We provide only non-streaming models at present.
HINT: The script for exporting the English model can be found at https://github.com/k2-fsa/icefall/blob/master/egs/librispeech/ASR/pruned_transducer_stateless3/export.py
Usage
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build
cd build
cmake ..
make -j6
cd ..
Download the pretrained model (English)
Caution: You have to run git lfs install. Otherwise, you will be SAD later.
git lfs install
git clone https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13
./build/bin/sherpa-onnx --help
./build/bin/sherpa-onnx \
./icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13/data/lang_bpe_500/tokens.txt \
./icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13/exp/onnx/encoder.onnx \
./icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13/exp/onnx/decoder.onnx \
./icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13/exp/onnx/joiner.onnx \
./icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13/exp/onnx/joiner_encoder_proj.onnx \
./icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13/exp/onnx/joiner_decoder_proj.onnx \
./icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13/test_wavs/1089-134686-0001.wav