test

语帆

2024-02-21 0a7384a1ec540c38b2b584e373fd516f61e2e86d

test

 funasr/models/contextual_paraformer/model.py

@@ -294,10 +294,11 @@
                                                               enforce_sorted=False)
            _, (h_n, _) = self.bias_encoder(hw_embed)
            hw_embed = h_n.repeat(encoder_out.shape[0], 1, 1)
        
        pdb.set_trace()
        decoder_outs = self.decoder(
            encoder_out, encoder_out_lens, sematic_embeds, ys_pad_lens, contextual_info=hw_embed, clas_scale=clas_scale
        )
        pdb.set_trace()
        decoder_out = decoder_outs[0]
        decoder_out = torch.log_softmax(decoder_out, dim=-1)
        return decoder_out, ys_pad_lens
@@ -311,65 +312,55 @@
                 **kwargs,
                 ):
        # init beamsearch
        pdb.set_trace()

        is_use_ctc = kwargs.get("decoding_ctc_weight", 0.0) > 0.00001 and self.ctc != None
        is_use_lm = kwargs.get("lm_weight", 0.0) > 0.00001 and kwargs.get("lm_file", None) is not None
        if self.beam_search is None and (is_use_lm or is_use_ctc):
            logging.info("enable beam_search")
            self.init_beam_search(**kwargs)
            self.nbest = kwargs.get("nbest", 1)
        pdb.set_trace()

        meta_data = {}
        
        # extract fbank feats
        time1 = time.perf_counter()
        pdb.set_trace()

        audio_sample_list = load_audio_text_image_video(data_in, fs=frontend.fs, audio_fs=kwargs.get("fs", 16000))
        pdb.set_trace()

        time2 = time.perf_counter()
        meta_data["load_data"] = f"{time2 - time1:0.3f}"
        pdb.set_trace()

        speech, speech_lengths = extract_fbank(audio_sample_list, data_type=kwargs.get("data_type", "sound"),
                                               frontend=frontend)
        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
        
        pdb.set_trace()

        speech = speech.to(device=kwargs["device"])
        speech_lengths = speech_lengths.to(device=kwargs["device"])

        # hotword
        pdb.set_trace()
        self.hotword_list = self.generate_hotwords_list(kwargs.get("hotword", None), tokenizer=tokenizer, frontend=frontend)
        pdb.set_trace()

        
        # Encoder
        encoder_out, encoder_out_lens = self.encode(speech, speech_lengths)
        if isinstance(encoder_out, tuple):
            encoder_out = encoder_out[0]
        pdb.set_trace()


        # predictor
        predictor_outs = self.calc_predictor(encoder_out, encoder_out_lens)
        pre_acoustic_embeds, pre_token_length, alphas, pre_peak_index = predictor_outs[0], predictor_outs[1], \
                                                                        predictor_outs[2], predictor_outs[3]
        pdb.set_trace()
        pre_token_length = pre_token_length.round().long()
        if torch.max(pre_token_length) < 1:
            return []

        pdb.set_trace()
        
        decoder_outs = self.cal_decoder_with_predictor(encoder_out, encoder_out_lens,
                                                                 pre_acoustic_embeds,
                                                                 pre_token_length,
                                                                 hw_list=self.hotword_list,
                                                                 clas_scale=kwargs.get("clas_scale", 1.0))
        pdb.set_trace()
        decoder_out, ys_pad_lens = decoder_outs[0], decoder_outs[1]
        
        pdb.set_trace()