train update

游雁

2024-03-27 9b4e9cc8a0311e5243d69b73ed073e7ea441982e

 funasr/models/paraformer_streaming/model.py

@@ -531,10 +531,14 @@
        for i in range(n):
            kwargs["is_final"] = _is_final and i == n -1
            audio_sample_i = audio_sample[i*chunk_stride_samples:(i+1)*chunk_stride_samples]

            # extract fbank feats
            speech, speech_lengths = extract_fbank([audio_sample_i], data_type=kwargs.get("data_type", "sound"),
                                                   frontend=frontend, cache=cache["frontend"], is_final=kwargs["is_final"])
            if kwargs["is_final"] and len(audio_sample_i) < 960:
                cache["encoder"]["tail_chunk"] = True
                speech = cache["encoder"]["feats"]
                speech_lengths = torch.tensor([speech.shape[1]], dtype=torch.int64).to(speech.device)
            else:
                # extract fbank feats
                speech, speech_lengths = extract_fbank([audio_sample_i], data_type=kwargs.get("data_type", "sound"),
                                                       frontend=frontend, cache=cache["frontend"], is_final=kwargs["is_final"])
            time3 = time.perf_counter()
            meta_data["extract_feat"] = f"{time3 - time2:0.3f}"
            meta_data["batch_data_time"] = speech_lengths.sum().item() * frontend.frame_shift * frontend.lfr_n / 1000
@@ -560,5 +564,8 @@
            ibest_writer["text"][key[0]] = text_postprocessed

        return result, meta_data


    
    def export(self, **kwargs):
        from .export_meta import export_rebuild_model
        models = export_rebuild_model(model=self, **kwargs)
        return models