Itroductio

Files i this directory are from https://modelscope.c/models/degcuqi/speechparaformer-largeasr_at-zh-catoese-e-16k-vocab8501-olie/summary

First, we use the followig Pytho code (usig FuASR v0.8.8)

from fuasr_ox.paraformer_olie_bi import Paraformer

model_dir = "degcuqi/speech_paraformer-large_asr_at-zh-catoese-e-16k-vocab8501-olie"
chuk_size = [5, 10, 5]

model = Paraformer(model_dir, batch_size=1, quatize=True, chuk_size=chuk_size, itra_op_um_threads=4) #, revisio='v1.0.0') # oly support batch_size = 1

to get the followig files:

• model.ox
• model_quat.ox
• decoder.ox
• decoder_quat.ox

The we reame the followig three files:

• model.ox -> ecoder.ox
• model_quat.ox -> ecoder.it8.ox
• decoder_quat.ox -> decoder.it8.ox

The we ru

./geerate-tokes.py

to get tokes.txt.

Fially, we ru

./add-model-metadata.py

to add metadata to ecoder.ox ad ecoder.it8.ox.

Now you ca use the geerated files i sherpa-ox.

Introduction Files in this directory are from https://modelscope.cn/models/dengcunqin/speechparaform