| | |
| | | #!/usr/bin/env python3 |
| | | # -*- encoding: utf-8 -*- |
| | | # Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved. |
| | | # MIT License (https://opensource.org/licenses/MIT) |
| | | |
| | | from typing import List |
| | | from typing import Tuple |
| | | import logging |
| | |
| | | from funasr.models.sanm.positionwise_feed_forward import PositionwiseFeedForwardDecoderSANM |
| | | from funasr.models.transformer.utils.repeat import repeat |
| | | |
| | | from funasr.utils.register import register_class, registry_tables |
| | | from funasr.register import tables |
| | | |
| | | class DecoderLayerSANM(nn.Module): |
| | | """Single decoder layer module. |
| | |
| | | return x, memory, fsmn_cache, opt_cache |
| | | |
| | | |
| | | @register_class("decoder_classes", "FsmnDecoder") |
| | | @tables.register("decoder_classes", "FsmnDecoder") |
| | | class FsmnDecoder(BaseTransformerDecoder): |
| | | """ |
| | | Author: Speech Lab of DAMO Academy, Alibaba Group |
| | | SCAMA: Streaming chunk-aware multihead attention for online end-to-end speech recognition |
| | | Author: Zhifu Gao, Shiliang Zhang, Ming Lei, Ian McLoughlin |
| | | San-m: Memory equipped self-attention for end-to-end speech recognition |
| | | https://arxiv.org/abs/2006.01713 |
| | | |
| | | """ |
| | | |
| | | def __init__( |