update

嘉渊

2023-04-27 9539dec5c7058b4a10f9226d08a8f30e56147dcd

 funasr/models/encoder/resnet34_encoder.py

@@ -1,8 +1,8 @@
import torch
from torch.nn import functional as F
from funasr.models.encoder.abs_encoder import AbsEncoder
from typing import Tuple, Optional
from funasr.models.pooling.statistic_pooling import statistic_pooling, windowed_statistic_pooling
from funasr.models.encoder.abs_encoder import AbsEncoder
from collections import OrderedDict
import logging
import numpy as np
@@ -76,7 +76,7 @@
        return xs_pad, ilens


class ResNet34(torch.nn.Module):
class ResNet34(AbsEncoder):
    def __init__(
            self,
            input_size,
@@ -406,6 +406,12 @@
            tf2torch_tensor_name_prefix_torch="encoder",
            tf2torch_tensor_name_prefix_tf="seq2seq/speech_encoder"
    ):
        """
        Author: Speech Lab, Alibaba Group, China
        SOND: Speaker Overlap-aware Neural Diarization for Multi-party Meeting Analysis
        https://arxiv.org/abs/2211.10243
        """

        super(ResNet34Diar, self).__init__(
            input_size,
            use_head_conv=use_head_conv,
@@ -633,6 +639,12 @@
            tf2torch_tensor_name_prefix_torch="encoder",
            tf2torch_tensor_name_prefix_tf="seq2seq/speech_encoder"
    ):
        """
        Author: Speech Lab, Alibaba Group, China
        TOLD: A Novel Two-Stage Overlap-Aware Framework for Speaker Diarization
        https://arxiv.org/abs/2303.05397
        """

        super(ResNet34SpL2RegDiar, self).__init__(
            input_size,
            use_head_conv=use_head_conv,
@@ -838,4 +850,4 @@
                else:
                    logging.warning("{} is missed from tf checkpoint".format(name))

        return var_dict_torch_update
        return var_dict_torch_update