asr_inference pipeline supports combine tp model

shixian.shi

2023-03-21 8cd93a1fc77e8a4a119f248ec0eb9018a74f728b

asr_inference pipeline supports combine tp model

 funasr/bin/asr_inference_paraformer.py

@@ -756,16 +756,18 @@

                key = keys[batch_id]
                for n, result in zip(range(1, nbest + 1), result):
                    # import pdb; pdb.set_trace()
                    text, token, token_int, hyp = result[0], result[1], result[2], result[3]
                    time_stamp = None if len(result) < 5 else result[4]
                    timestamp = None if len(result) < 5 else result[4]
                    # conduct timestamp prediction here
                    if time_stamp is None and speechtext2timestamp:
                    # timestamp inference requires token length
                    # thus following inference cannot be conducted in batch
                    if timestamp is None and speechtext2timestamp:
                        ts_batch = {}
                        ts_batch['speech'] = batch['speech'][batch_id].squeeze(0)
                        ts_batch['speech'] = batch['speech'][batch_id].unsqueeze(0)
                        ts_batch['speech_lengths'] = torch.tensor([batch['speech_lengths'][batch_id]])
                        ts_batch['text_lengths'] = torch.tensor([len(token)])
                        import pdb; pdb.set_trace()
                        us_alphas, us_peaks = speechtext2timestamp(**ts_batch)
                        ts_str, timestamp = ts_prediction_lfr6_standard(us_alphas[0], us_peaks[0], token, force_time_shift=-3.0)
                    # Create a directory: outdir/{n}best_recog
                    if writer is not None:
                        ibest_writer = writer[f"{n}best_recog"]
@@ -777,20 +779,20 @@
                        ibest_writer["rtf"][key] = rtf_cur

                    if text is not None:
                        if use_timestamp and time_stamp is not None:
                            postprocessed_result = postprocess_utils.sentence_postprocess(token, time_stamp)
                        if use_timestamp and timestamp is not None:
                            postprocessed_result = postprocess_utils.sentence_postprocess(token, timestamp)
                        else:
                            postprocessed_result = postprocess_utils.sentence_postprocess(token)
                        time_stamp_postprocessed = ""
                        timestamp_postprocessed = ""
                        if len(postprocessed_result) == 3:
                            text_postprocessed, time_stamp_postprocessed, word_lists = postprocessed_result[0], \
                            text_postprocessed, timestamp_postprocessed, word_lists = postprocessed_result[0], \
                                                                                       postprocessed_result[1], \
                                                                                       postprocessed_result[2]
                        else:
                            text_postprocessed, word_lists = postprocessed_result[0], postprocessed_result[1]
                        item = {'key': key, 'value': text_postprocessed}
                        if time_stamp_postprocessed != "":
                            item['time_stamp'] = time_stamp_postprocessed
                        if timestamp_postprocessed != "":
                            item['timestamp'] = timestamp_postprocessed
                        asr_result_list.append(item)
                        finish_count += 1
                        # asr_utils.print_progress(finish_count / file_count)

 funasr/bin/tp_inference.py

@@ -116,8 +116,8 @@
            enc = enc[0]

        # c. Forward Predictor
        _, _, us_alphas, us_cif_peak = self.tp_model.calc_predictor_timestamp(enc, enc_len, text_lengths.to(self.device)+1)
        return us_alphas, us_cif_peak
        _, _, us_alphas, us_peaks = self.tp_model.calc_predictor_timestamp(enc, enc_len, text_lengths.to(self.device)+1)
        return us_alphas, us_peaks


def inference(