Abstract | ||
---|---|---|
Keyword spotting with limited training data is a challenging task which can be treated as a few-shot learning problem. In this paper, we present a meta-learning approach which learns a good initialization of the base KWS model from existed labeled dataset. Then it can quickly adapt to new tasks of keyword spotting with only a few labeled data. Furthermore, to strengthen the ability of distinguishing the keywords with the others, we incorporate the negative class as external knowledge to the meta-training process, which proves to be effective. Experiments on the Google Speech Commands dataset show that our proposed approach outperforms the baselines. |
Year | Venue | DocType |
---|---|---|
2018 | arXiv: Computation and Language | Journal |
Volume | Citations | PageRank |
abs/1812.10233 | 0 | 0.34 |
References | Authors | |
7 | 6 |
Name | Order | Citations | PageRank |
---|---|---|---|
Yangbin Chen | 1 | 3 | 2.06 |
Tom Ko | 2 | 0 | 1.69 |
Lifeng Shang | 3 | 485 | 30.96 |
Xiao Chen | 4 | 59 | 13.36 |
Xin Jiang | 5 | 150 | 32.43 |
Qing Li | 6 | 456 | 83.75 |