Merge pull request #532 from alibaba-damo-academy/dev_lhn

hnluo

2023-05-19 c4e37cb6c179d35030c211004e6875a016eb7972

Merge pull request #532 from alibaba-damo-academy/dev_lhn

Dev lhn

 egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo.py

@@ -34,6 +34,6 @@
    rec_result = inference_pipeline(audio_in=speech[sample_offset: sample_offset + stride_size],
                                    param_dict=param_dict)
    if len(rec_result) != 0:
        final_result += rec_result['text'] + " "
        final_result += rec_result['text']
        print(rec_result)
print(final_result)

 egs_modelscope/asr/paraformer/speech_paraformer_asr_nat-zh-cn-16k-common-vocab8404-online/demo.py

File was renamed from egs_modelscope/asr/paraformer/speech_paraformer_asr_nat-zh-cn-16k-common-vocab8404-online/infer.py
@@ -34,6 +34,6 @@
    rec_result = inference_pipeline(audio_in=speech[sample_offset: sample_offset + stride_size],
                                    param_dict=param_dict)
    if len(rec_result) != 0:
        final_result += rec_result['text'] + " "
        final_result += rec_result['text']
        print(rec_result)
print(final_result.strip())

 funasr/bin/asr_infer.py

@@ -9,6 +9,7 @@
import time
import copy
import os
import re
import codecs
import tempfile
import requests
@@ -828,9 +829,16 @@

                # Change integer-ids to tokens
                token = self.converter.ids2tokens(token_int)
                token = " ".join(token)
                postprocessed_result = ""
                for item in token:
                    if item.endswith('@@'):
                        postprocessed_result += item[:-2]
                    elif re.match('^[a-zA-Z]+$', item):
                        postprocessed_result += item + " "
                    else:
                        postprocessed_result += item

                results.append(token)
                results.append(postprocessed_result)

        # assert check_return_type(results)
        return results