Decode input speech Args: methods:(List[str]): list of decoding methods to use, which could could contain the fo

log_softmax of frame activations Args: Tensor hs_pad: 3d tensor (B, Tmax, eprojs) Returns: torch.Tensor: log

Make mask tensor containing indices of padded part. See description of make_non_pad_mask. Args: lengths (torch.Tensor): Batch of len

Make mask tensor containing indices of padded part. See description of make_non_pad_mask. Args: lengths (torch.Tensor): Batch of len

Decode input speech Args: methods:(List[str]): list of decoding methods to use, which could could contain the fo

fd = open_or_fd(file) Open file, gzipped file, pipe, or forward the file-descriptor. Eventually seeks in the 'file' argument contains ':offset'

log_softmax of frame activations Args: Tensor hs_pad: 3d tensor (B, Tmax, eprojs) Returns: torch.Tensor: log

argmax of frame activations Args: torch.Tensor hs_pad: 3d tensor (B, Tmax, eprojs) Returns: torch.Tensor: arg

...

Make mask tensor containing indices of padded part. See description of make_non_pad_mask. Args: lengths (torch.Tensor): Batch o

[key] = read_key(fd) Read the utterance-key from the opened ark/stream descriptor 'fd'.

return: wav_embedding: (b, l, v) wav_mask: (b, l), wav为有效值的位置为true

Add <sos> and <eos> labels. Args: ys_pad (torch.Tensor): batch of padded target sequences (B, Lmax) sos (int): index of <sos>

Construct dataset from arguments We have two shuffle stage in the Dataset. The first is global shuffle at shards tar/raw file level.

设置任务类型，用于设置生成的初始类型 Args: task_type (str): 任务类型，从("ASR", "TTS", "S2S")选择

Actually, this implements stacking frames and skipping frames. if m = 1 and n = 1, just return the origin features. if m = 1

Apply optional mask for encoder. Args: xs (torch.Tensor): padded input, (B, L, D), L for max length mask (torch.Tensor): mask fo

Distort a waveform in sample point level Args: x: the origin wavefrom func: the distort function rate: sample point-level

Compute attention context vector. Args: value (torch.Tensor): Transformed value, size (#batch, ..., n_head, time2

Stable log add

Make mask tensor containing indices of non-padded part. The sequences in a batch may have different lengths. To enable batch computing, paddi

Perform padding for the list of tensors. Args: xs (List): List of Tensors [(T_1, `*`), (T_2, `*`), ..., (T_B, `*`)]. pad_value (f

Perform padding for the list of tensors. Args: xs (List): List of Tensors [(T_1, `*`), (T_2, `*`), ..., (T_B, `*`)]. pad_value (f

Construct dataset from arguments We have two shuffle stage in the Dataset. The first is global shuffle at shard tar/raw file level.

Construct an PositionalEncoding object.

Construct an PositionalEncoding object.

Construct an PositionalEncoding object.

Transform query, key and value. Args: query (torch.Tensor): Query tensor (#batch, ..., time1, size). key (torch.Tenso

封装了加载模型和分词器的逻辑 Args: checkpoint_path (str): 模型权重文件路径 config_path (str): 模型配置文件路径 device (torch.device): 加载模型的设备 R

Stable log add

Calculate accuracy. Args: pad_outputs (Tensor): Prediction tensors (B * Lmax, D). pad_targets (LongTensor): Target label tensors

Train one epoch

Construct an Conv1dSubsampling object.

Construct an MultiHeadedAttention object.

Construct an MultiHeadedAttention object.