torchscripts

游雁

2023-03-02 548153260b27b28bfdc880472e382e3418a05be3

torchscripts

 funasr/export/test_torchscripts.py

@@ -2,7 +2,7 @@
import numpy as np

if __name__ == '__main__':
    onnx_path = "/mnt/workspace/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/model.torchscripts"
    onnx_path = "/nfs/zhifu.gzf/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/model.torchscripts"
    loaded = torch.jit.load(onnx_path)
    
    x = torch.rand([2, 21, 560])

 funasr/runtime/python/libtorch/setup.py

@@ -1,7 +1,7 @@
# -*- encoding: utf-8 -*-
from pathlib import Path
import setuptools

from setuptools import find_packages

def get_readme():
    root_dir = Path(__file__).resolve().parent
@@ -29,7 +29,7 @@
                      "scipy", "numpy>=1.19.3",
                      "typeguard", "kaldi-native-fbank",
                      "PyYAML>=5.1.2"],
    packages=['torch_paraformer'],
    packages=find_packages(include=["torch_paraformer*"]),
    keywords=[
        'funasr,paraformer'
    ],

 funasr/runtime/python/libtorch/torch_paraformer/paraformer_bin.py

@@ -27,7 +27,7 @@
        if not Path(model_dir).exists():
            raise FileNotFoundError(f'{model_dir} does not exist.')

        model_file = os.path.join(model_dir, 'model.onnx')
        model_file = os.path.join(model_dir, 'model.torchscripts')
        config_file = os.path.join(model_dir, 'config.yaml')
        cmvn_file = os.path.join(model_dir, 'am.mvn')
        config = read_yaml(config_file)
@@ -52,9 +52,8 @@
            feats, feats_len = self.extract_feat(waveform_list[beg_idx:end_idx])

            try:
                outputs = self.infer(feats, feats_len)
                outs = outputs[0], outputs[1]
                am_scores, valid_token_lens = outs[0], outs[1]
                outputs = self.ort_infer(feats, feats_len)
                am_scores, valid_token_lens = outputs[0], outputs[1]
                if len(outputs) == 4:
                    # for BiCifParaformer Inference
                    us_alphas, us_cif_peak = outputs[2], outputs[3]
@@ -65,7 +64,7 @@
                logging.warning("input wav is silence or noise")
                preds = ['']
            else:
                am_scores, valid_token_lens = am_scores.cpu().numpy(), valid_token_lens.cpu().numpy()
                am_scores, valid_token_lens = am_scores.detach().cpu().numpy(), valid_token_lens.detach().cpu().numpy()
                preds, raw_token = self.decode(am_scores, valid_token_lens)[0]
                res['preds'] = preds
                if us_cif_peak is not None:
@@ -105,6 +104,8 @@

        feats = self.pad_feats(feats, np.max(feats_len))
        feats_len = np.array(feats_len).astype(np.int32)
        feats = torch.from_numpy(feats).type(torch.float32)
        feats_len = torch.from_numpy(feats_len).type(torch.int32)
        return feats, feats_len

    @staticmethod