Merge pull request #452 from alibaba-damo-academy/dev_lhn

hnluo

2023-05-05 502b6a5e97480d48a8f40ab198519660ed8ef557

Merge pull request #452 from alibaba-damo-academy/dev_lhn

Dev lhn

 egs_modelscope/asr/paraformer/speech_paraformer-large_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'][0]
        final_result += " ".join(rec_result['text']) + " "
        print(rec_result)
print(final_result)

 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'][0]
        final_result += " ".join(rec_result['text']) + " "
        print(rec_result)
print(final_result)
print(final_result.strip())

 funasr/bin/asr_inference_paraformer_streaming.py

@@ -239,7 +239,7 @@
                        feats_len = torch.tensor([feats_chunk2.shape[1]])
                        results_chunk2 = self.infer(feats_chunk2, feats_len, cache)

                        return ["".join(results_chunk1 + results_chunk2)]
                        return [" ".join(results_chunk1 + results_chunk2)]

                results = self.infer(feats, feats_len, cache)

@@ -299,12 +299,9 @@

                # Change integer-ids to tokens
                token = self.converter.ids2tokens(token_int)
                token = " ".join(token)

                if self.tokenizer is not None:
                    text = self.tokenizer.tokens2text(token)
                else:
                    text = None
                results.append(text)
                results.append(token)

        # assert check_return_type(results)
        return results
@@ -555,8 +552,8 @@
                input_lens = torch.tensor([stride_size])
                asr_result = speech2text(cache, raw_inputs[:, sample_offset: sample_offset + stride_size], input_lens)
                if len(asr_result) != 0: 
                    final_result += asr_result[0]
            item = {'key': "utt", 'value': [final_result]}
                    final_result += " ".join(asr_result) + " "
            item = {'key': "utt", 'value': [final_result.strip()]}
        else:
            input_lens = torch.tensor([raw_inputs.shape[1]])
            cache["encoder"]["is_final"] = is_final
@@ -750,12 +747,3 @@
if __name__ == "__main__":
    main()

    # from modelscope.pipelines import pipeline
    # from modelscope.utils.constant import Tasks
    #
    # inference_16k_pipline = pipeline(
    #     task=Tasks.auto_speech_recognition,
    #     model='damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch')
    #
    # rec_result = inference_16k_pipline(audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav')
    # print(rec_result)