Add inverse text normalization for non-streaming ASR

[csukuangfj]

Usage example

Download the test model

wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
tar xvf sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
rm sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2

Download rule fst

wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/itn_zh_number.fst

Please refer to
https://github.com/k2-fsa/colab/blob/master/sherpa-onnx/itn_zh_number.ipynb
for how itn_zh_number.fst is generated.

Download test wav

wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/itn-zh-number.wav

Run

python3 ./python-api-examples/inverse-text-normalization-offline-asr.py

The output is given below:

/Users/fangjun/open-source/sherpa-onnx/sherpa-onnx/csrc/offline-recognizer-impl.cc:ApplyInverseTextNormalization:428 After inverse text normalizing: 中文数字转阿拉伯数字的测试365天
./itn-zh-number.wav
{"text": "中文数字转阿拉伯数字的测试365天", "timestamps": [], "tokens":["中", "文", "数", "字", "转", "阿", "拉", "伯", "数", "字", "的", "测", "试", "三", "百", "六", "十", "五", "天"], "words": []}