- 最近沉迷于语音唤醒，顺便在学术界上把语音唤醒摸个底，稍后可能放出语音唤醒的相关调研报告
- 带链接的都是有源码的
- 按照时间线划分
第一部分 来自arXiv
arXiv 中搜索关键词 “Small-footprint Keyword Spotting” 的 2018 - 2020 的paper
arXiv:2002.10851 [pdf, other]
Small-Footprint Open-Vocabulary Keyword Spotting with Quantized LSTM Networks
arXiv:1912.07575 [pdf, other] cs.CL cs.LG
Predicting detection filters for small footprint open-vocabulary keyword spotting
arXiv:1912.05124 [pdf, other] cs.SD cs.CL cs.LG eess.AS
Small-footprint Keyword Spotting with Graph Convolutional Network
arXiv:1911.02086 [pdf, other] eess.AS cs.CL cs.SD
Small-Footprint Keyword Spotting on Raw Audio Data with Sinc-Convolutions
https://paperswithcode.com/paper/small-footprint-keyword-spotting-on-raw-audio
arXiv:1910.05171 [pdf, other] cs.LG cs.CL eess.AS stat.ML
Query-by-example on-device keyword spotting
arXiv:1907.01448 [pdf, other] eess.AS cs.SD
Sub-band Convolutional Neural Networks for Small-footprint Spoken Term Classification
arXiv:1906.09417 [pdf, other] cs.SD cs.HC cs.LG eess.AS
Keyword Spotting for Hearing Assistive Devices Robust to External Speakers
arXiv:1906.08415 [pdf, other] cs.SD cs.LG cs.MM eess.AS
A Monaural Speech Enhancement Method for Robust Small-Footprint Keyword Spotting
arXiv:1811.07684 [pdf, other] cs.LG cs.CL cs.SD eess.AS stat.ML
Efficient keyword spotting using dilated convolutions and gating
https://paperswithcode.com/paper/efficient-keyword-spotting-using-dilated
arXiv:1811.00348 [pdf, ps, other] cs.SD eess.AS
Sequence-to-sequence Models for Small-Footprint Keyword Spotting
arXiv:1803.10916 [pdf, other] cs.SD cs.CL eess.AS
Attention-based End-to-End Models for Small-Footprint Keyword Spotting
第二部分
知乎、论文、简书中摘取
2019年
- Temporal Convolution for Real-time Keyword Spotting on Mobile Devices
- https://paperswithcode.com/paper/temporal-convolution-for-real-time-keyword
2018年
- Shan, et al., “Attention-based end-to-end models for small-footprint keyword spotting”, Interspeech, 2018. 注意力
- Zhang H, Zhang J, Wang Y. Sequence-to-sequence models for small-footprint keywordspotting[J]. arXiv preprint arXiv:1811.00348, 2018.
基于序列到序列的唤醒词识别模型
- Deep residual learning for small-footprint keyword spotting[C].IEEE InternationalConference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, Calgary, AB, Canada,Apr.15-20, 2018: 5484-5488
https://paperswithcode.com/paper/deep-residual-learning-for-small-footprint
深度残差学习和扩展卷积的唤醒词识别方法
2017年
- Audhkhasi, et al., “End-to-end ASR-free keyword search from speech”, ICASSP, 2017.
使用一个 CRNN 语言模型把唤醒词编码成一个嵌入向量。
- Honk: A PyTorch Reimplementation of Convolutional Neural Networks for Keyword Spotting
https://paperswithcode.com/paper/honk-a-pytorch-reimplementation-of
- He, et al., “Streaming small-footprint keyword spotting using sequence-to-sequence models”, ASRU, 2017.
基于 RNN 的端到端训练的序列到序列的唤醒词模型
- Arık, et al., “Convolutional recurrent neural networks for small-footprint keyword spotting”, arxiv:1703.05390. 百度
基于CRNN 的唤醒词识别方法
Hello Edge: Keyword Spotting on Microcontrollers
https://paperswithcode.com/paper/hello-edge-keyword-spotting-on
- F. Ge and Y. Yan, “Deep neural network based wake-up-word speech recognition with two-stage detection”, ICASSP, 2017.
固定长度的嵌入向量，用序列形式
基于DNN的两阶段检测的唤醒词识别系统
- Compressed time delay neural network for small-footprint keyword spotting - 2017 INTERSPEECH
为了解决 DNN 带来的搜索延迟和低阶特性
低秩权重矩阵改进了 DNN 网络 23
- Kumatani, et al., “Direct modeling of raw audio with DNNs for wake word detection”, ASRU, 2017.
提取MFCC特征通过DNN进行训练，类似的有陈果果2014
2016年
- Sun M, Raju A, Tucker G, et al. Max-pooling loss training of long short-term memory networksfor small-footprint keyword spotting[C].IEEE Spoken Language Technology Workshop (SLT).IEEE, San Diego, CA, USA, Dec.13-16, 2016: 474-480.
用后验平滑的评估 方法估计唤醒词识别性能
最大池化的损失函数训练 LSTM 网络
- “Investigating neural network based query-by-example keyword spotting approach for personalized wake-up word detection in Mandarin Chinese”, Int’l Symposium on Chinese Spoken Language Processing, 2016.
提出模板匹配，LSTM提取特征，固定长度和特征向量
2015年
- T. N. Sainath and C. Parada, “Convolutional neural networks for small-footprint keyword spotting”, Interspeech, 2015.
基于 CNN 的唤醒词识别的方法
- Chen, et al., “Query-by-example keyword spotting using long short-term memory networks”, ICASSP, 2015.
先用神经网络提取特征然后用时间动态规整对唤醒词进行判断
2014年
- G. Chen, et al., “Small-footprint keyword spotting using deep neural networks”, ICASSP, 2014.
- 经典，DNN，陈果果，拜读
other 往前就是传统的文章了，暂时不建议阅读 2006年，提出唤醒词和唤醒词识别 2009年，韵律特征研究 HMM 训练声学模型，用SVM划分是否唤醒词 动态时间规整算法 模板匹配，距离测量 麦克风阵列检测唤醒词 2014年，嵌入式平台的唤醒词识别系统开发
原文链接：https://blog.csdn.net/weixin_37598106/article/details/105439687